Editors: Heather Tallis, Taylor Ricketts, Anne Guerry, Spencer Wood, and Richard Sharp. Contributing Authors: Erik Nelson, Driss Ennaanay, Stacie Wolny, Nasser Olwero, Kari Vigerstol, Derric Penning- ton, Guillermo Mendoza, Juliann Aukema, John Foster, Jessica Forrest, Dick Cameron, Katie Arkema, Eric Lonsdorf, Christina Kennedy, Gregory Verutes, Chong-Ki Kim, Gregory Guannel, Michael Papenfus, Jodie Toft, Matthew Mar- sik, Joey Bernhardt, Robert Griffin, Kathryn Glowinski, Nicholas Chaumont, Adam Perelman, and Martin Lacayo. . Citation: Tallis, H.T., Ricketts, T., Guerry, A.D., Wood, S.A., Sharp, R., Nelson, E., Ennaanay, D., Wolny, S., Olwero, N., Vigerstol, K., Pennington, D., Mendoza, G., Aukema, J., Foster, J., Forrest, J., Cameron, D., Arkema, K., Lonsdorf, E., Kennedy, C., Verutes, G., Kim, C.K., Guannel, G., Papenfus, M., Toft, J., Marsik, M., Bernhardt, J., and Griffin, R., Glowinski, K., Chaumont, N., Perelman, A., Lacayo, M. 2013. InVEST 2.5.6 User’s Guide. The Natural Capital Project, Stanford.
i CONTENTS
I Front-matter1
1 Data Requirements and Outputs Summary Table2
2 The Need for a New Tool 8 2.1 Introduction...... 8 2.2 Who should use InVEST?...... 8 2.3 Introduction to InVEST...... 9 2.4 A work in progress...... 11 2.5 This guide...... 11
3 Getting Started 13 3.1 Getting started with InVEST...... 13 3.2 Installing the InVEST tool and data on your computer...... 13 3.3 Downloading and installing Python library extensions...... 13 3.4 Adding the InVEST toolbox to ArcMap...... 14 3.5 Using sample data...... 15 3.6 Formatting your data...... 15 3.7 Running the models...... 16 3.8 Support information...... 17 3.9 Model run checklist...... 19 3.10 Reporting errors...... 19 3.11 Working with the DEM...... 19 3.12 Resources...... 22
II Marine Models 23
4 Wave Energy Model 24 4.1 Summary...... 24 4.2 Introduction...... 24 4.3 The model...... 25 4.4 Data needs...... 29 4.5 Running the model...... 32 4.6 Interpreting results...... 35 4.7 Case example illustrating results...... 38 4.8 Appendix A...... 41 4.9 Wave Energy 3.0 Beta...... 41 4.10 References...... 42
ii 5 Wind Energy Model 43 5.1 Summary...... 43 5.2 Introduction...... 43 5.3 The model...... 44 5.4 Data Needs...... 53 5.5 Interpreting Results...... 54 5.6 Data Sources...... 55 5.7 Running the Model...... 55 5.8 References...... 56
6 Coastal Vulnerability Model 57 6.1 Summary...... 57 6.2 Introduction...... 57 6.3 The model...... 58 6.4 Limitations and Simplifications...... 64 6.5 Data needs...... 64 6.6 Running the model...... 70 6.7 Interpreting results...... 74 6.8 Appendix A...... 76 6.9 Appendix B...... 76 6.10 References...... 82
7 Erosion Protection Model 84 7.1 Summary...... 84 7.2 Introduction...... 84 7.3 The model...... 85 7.4 Limitations and Simplifications...... 97 7.5 Data Needs...... 99 7.6 Running the model...... 114 7.7 Interpreting results...... 121 7.8 References...... 126 7.9 Appendix A...... 127
8 Marine Fish Aquaculture 132 8.1 Summary...... 132 8.2 Introduction...... 132 8.3 The Model...... 133 8.4 Limitations and simplifications...... 135 8.5 Data needs...... 136 8.6 Running the model...... 138 8.7 Interpreting results...... 142 8.8 References...... 145
9 Aesthetic Quality 146 9.1 Summary...... 146 9.2 Introduction...... 146 9.3 The model...... 147 9.4 Limitations and simplifications...... 148 9.5 Data needs...... 148 9.6 Running the model...... 149 9.7 Interpreting results...... 154 9.8 Case example illustrating results...... 155 9.9 References...... 159
10 Overlap Analysis Model 160
iii 10.1 Summary...... 160 10.2 Introduction...... 160 10.3 The model...... 161 10.4 Limitations and simplifications...... 162 10.5 Data needs...... 163 10.6 Interpreting results...... 166 10.7 Appendix A...... 167 10.8 References...... 169
11 Habitat Risk Assessment 170 11.1 Summary...... 170 11.2 Introduction...... 170 11.3 The model...... 172 11.4 Data Needs...... 179 11.5 Interpreting results...... 186 11.6 References...... 190
12 Marine Water Quality Model: Advection-Diffusion Model 191 12.1 Summary...... 191 12.2 Introduction...... 191 12.3 The model...... 192 12.4 Limitations and simplifications...... 195 12.5 Data Needs...... 195 12.6 Running the Model...... 196 12.7 Interpreting Results...... 196 12.8 Case example illustrating model inputs and results...... 196 12.9 References...... 200
III Terrestrial and Freshwater Models 201
13 Biodiversity: Habitat Quality & Rarity 202 13.1 Summary...... 202 13.2 Introduction...... 202 13.3 The Model...... 203 13.4 Data needs...... 209 13.5 Running the Model...... 212 13.6 Interpreting Results...... 214 13.7 Biodiversity 3.0 Beta...... 216 13.8 References...... 216
14 Carbon Storage and Sequestration 218 14.1 Summary...... 218 14.2 Introduction...... 218 14.3 The Model...... 219 14.4 Data needs...... 224 14.5 Running the Model...... 232 14.6 Interpreting Results...... 234 14.7 Appendix: data sources...... 236 14.8 Carbon 3.0 Beta...... 241 14.9 References...... 241
15 Reservoir Hydropower Production 246 15.1 Summary...... 246 15.2 Introduction...... 246
iv 15.3 The Model...... 247 15.4 Data needs...... 251 15.5 Running the Model...... 255 15.6 Interpreting Results...... 258 15.7 Appendix A: Data Sources...... 261 15.8 Appendix B: Calibration of Water Yield Model...... 266 15.9 References...... 267
16 Water Purification: Nutrient Retention 268 16.1 Summary...... 268 16.2 Introduction...... 268 16.3 The Model...... 269 16.4 Data Needs...... 272 16.5 Running the Model...... 275 16.6 Interpreting Results...... 276 16.7 Appendix: Data Sources...... 278 16.8 References...... 282
17 Sediment Retention Model: Avoided dredging and water quality regulation 283 17.1 Summary...... 283 17.2 Sediment Updates to InVEST 3.0...... 283 17.3 Introduction...... 284 17.4 The Model...... 284 17.5 Data needs...... 289 17.6 Running the Model...... 292 17.7 Interpreting Results...... 293 17.8 Appendix: data sources...... 297 17.9 References...... 301
18 Managed Timber Production Model 302 18.1 Summary...... 302 18.2 Introduction...... 302 18.3 The Model...... 302 18.4 Data needs...... 305 18.5 Running the Model...... 309 18.6 Interpreting results...... 312 18.7 Timber 3.0 Beta...... 312 18.8 References...... 312
19 Crop Pollination 313 19.1 Summary...... 313 19.2 Introduction...... 313 19.3 The Model...... 314 19.4 Data needs...... 317 19.5 Running the Model...... 319 19.6 Interpreting results...... 322 19.7 Appendix: Data sources...... 323 19.8 References...... 323
IV Experimental Models 325
20 Malaria Mosquito Habitat 326 20.1 Summary...... 326 20.2 Introduction...... 326
v 20.3 The model...... 326 20.4 Limitations and simplifications...... 327 20.5 Data Needs...... 327 20.6 Running the Model...... 327 20.7 Interpreting Results...... 328 20.8 References...... 328
V Acknowledgements 329
21 Marine Models 330 21.1 Acknowledgements...... 330
VI Roadmap 332
22 Marine Models 333 22.1 What’s coming up for Marine InVEST models?...... 333
vi Part I
Front-matter
1 CHAPTER ONE
DATA REQUIREMENTS AND OUTPUTS SUMMARY TABLE
2 InVEST User Guide, Release 2.5.6
3 InVEST User Guide, Release 2.5.6
4 InVEST User Guide, Release 2.5.6
5 InVEST User Guide, Release 2.5.6
6 InVEST User Guide, Release 2.5.6
7 CHAPTER TWO
THE NEED FOR A NEW TOOL
2.1 Introduction
Ecosystems, if properly managed, yield a flow of services that are vital to humanity, including the production of goods (e.g., food), life support processes (e.g., water purification), and life fulfilling conditions (e.g., beauty, recreation opportunities), and the conservation of options (e.g., genetic diversity for future use). Despite its importance, this natural capital is poorly understood, scarcely monitored, and—in many cases—undergoing rapid degradation and depletion. To better align ecosystem conservation with economic forces, the Natural Capital Project is developing models that quantify and map the values of environmental services. The modeling suite is best suited for analyses of multiple services and multiple objectives. The current models, which require relatively little data input, can identify areas where investment may enhance human well-being and nature. We are continuing development of the models and will release new, updated versions as they become available.
2.2 Who should use InVEST?
InVEST is designed to inform decisions about natural resource management. Decision-makers, from governments to non-profits to corporations, often manage lands and waters for multiple uses and inevitably must evaluate trade-offs among these uses; InVEST’s multi-service, modular design provides an effective tool for evaluating these trade-offs. For example, government agencies could use InVEST to help determine how to manage lands, coasts, and marine areas to provide an optimal mix of benefits to people or to help design permitting and mitigation programs that sustain nature’s benefits to society. Conservation organizations could use InVEST to better align their missions to protect biodiversity with activities that improve human livelihoods. Corporations, such as timber companies, renewable energy companies, and water utilities, could also use InVEST to decide how and where to invest in natural capital to ensure that their supply chains are preserved. InVEST can help answer questions like: • Where do environmental services originate and where are they consumed? • How does a proposed forestry management plan affect timber yields, biodiversity, water quality and recreation? • What kinds of coastal management and fishery policies will yield the best returns for sustainable fisheries, shoreline protection and recreation? • Which parts of a watershed provide the greatest carbon sequestration, biodiversity, and tourism values? • Where would reforestation achieve the greatest downstream water quality benefits while maintaining or mini- mizing losses in water flows? • How will climate change and population growth impact environmental services and biodiversity?
8 InVEST User Guide, Release 2.5.6
• What benefits does marine spatial planning provide to society in addition to food from fishing and aquaculture and secure locations for renewable energy facilities?
2.3 Introduction to InVEST
The InVEST toolset described in this guide includes models for quantifying, mapping, and valuing the benefits pro- vided by terrestrial, freshwater and marine systems. Specifically it includes models for: • Wave Energy • Coastal Vulnerability • Coastal Protection • Marine Fish Aquaculture • Marine Aesthetic Quality • Marine Overlap Analysis Model: Fisheries and Recreation • Marine Habitat Risk Assessment • Terrestrial Biodiversity: Habitat Quality and Rarity • Carbon Storage and Sequestration • Reservoir Hydropower Production • Water Purification: Nutrient Retention • Sediment Retention Model: Avoided Dredgin and Water Quality Regulation • Manage Timber Production • Crop Pollination To date, the marine and terrestrial/freshwater models are treated separately. The current version of InVEST presents the models for the two systems in turn. In future releases, models for the two systems will be more integrated. This will occur in two primary ways. First, some models will have improved flexibility to be applied in either terrestrial or marine systems (e.g. carbon storage and sequestration, biodiversity, recreation, aesthetic views). (The terrestrial biodiversity model can be applied, as is, to nearshore marine systems. Please see the marine habitat quality model chapter for a discussion of differences between the approaches and why there are two). Second, we are working to link freshwater and marine models so that effects of watershed activities on coastal and marine systems can be explored. Such linkages will be included in later releases of InVEST. InVEST is most effectively used within a decision-making process that starts with a series of stakeholder consultations (illustrated in Figure 1). Through discussion, questions of interest to policy makers, communities and conservation groups are identified. These questions may concern service delivery on a landscape today and how these services may be affected by new programs, policies, and conditions in the future. For questions regarding the future, stakeholders develop “scenarios” to explore the consequences of expected changes on natural resources. These scenarios typically include a map of future land use and land cover or, for the marine models, a map of future coastal and ocean uses and coastal/marine habitats. Following stakeholder consultations and scenario development, InVEST can estimate the amount and value of envi- ronmental services that are provided on the current landscape or under future scenarios. InVEST models are spatially- explicit, using maps as information sources and producing maps as outputs. InVEST returns results in either biophys- ical terms (e.g., tons of carbon sequestered) or economic terms (e.g., net present value of that sequestered carbon). The spatial resolution of analyses is also flexible, allowing users to address questions at the local, regional or global scale. InVEST results can be shared with the stakeholders and decision-makers who created the scenarios to inform upcoming decisions. Using InVEST in an iterative process, these stakeholders may choose to create new scenarios based on the information revealed by the models until suitable solutions for management action are identified.
2.3. Introduction to InVEST 9 InVEST User Guide, Release 2.5.6
Figure 2.1: Schematic of the decision-making process in which InVEST is embedded. Stakeholders create scenarios that are assessed for environmental service value by biophysical and economic models that produce several types of outputs.
InVEST has a tiered design. Tier 0 models map relative levels of environmental services and/or highlight regions where particular services are in high demand. For example, the coastal vulnerability model in InVEST maps regions of the coastline that are particularly susceptible to erosion and flooding. It does not use a production function to yield outputs of meters of shoreline eroded or to value coastal protection services provided by nearshore marine habitats. There is no valuation done in tier 0 models. Tier 1 models are theoretically grounded but simple. They are suitable when more data are available than are required for Tier 0, but they still have relatively simple data requirements. Tier 1 models can identify areas of high or low environmental service production and biodiversity across the landscape, and the tradeoffs and synergies among services under current or future conditions. Tier 1 models give outputs in absolute terms and provide the option for economic valuation (except for biodiversity). For example, the Finfish Aquaculture model can provide outputs in lbs. of fish or in $’s. More complex Tier 2 models are under development for biodiversity and some environmental services. Tier 2 mod- els provide increasingly precise estimates of environmental services and values, which are especially important for establishing contracts for payments for environmental services programs or assessing scenarios that address change on a sub-annual basis. For example, scenarios that represent a change in the monthly or seasonal timing of fertilizer application or water extraction in agricultural systems cannot be assessed by Tier 1 models, but will be treated well by Tier 2 models. We expect users to mix and match Tier 0, 1 and (later) 2 models to create the best suite of models given past work, existing data, and the questions of interest. Although the more sophisticated models require substantial data and time to develop, once they are parameterized for a certain location, they will provide the best available science for new decisions. In some cases (e.g. for fisheries), complex tier 3-type models already exist in a particular location. The Natural Capital Project will not develop new tier 3 models, but rather sees these as the sophisticated, dynamic models usually developed for individual sites or contexts. We aim to develop the capability of InVEST to communicate with such existing, complex models so that InVEST inputs (e.g. scenarios) can be fed in, and outputs from those complex models can be compared with other InVEST outputs. InVEST includes a mixture of Tier 0 and Tier 1 models. Tier 2 models for several services have been formulated and documented in Natural Capital: The Theory & Practice of Mapping Environmental Services, a book just published by Oxford University Press. We will design the Tier 2 software platform as a space where Tier 0, 1, 2 and 3 models can be integrated as appropriate for different applications.
2.3. Introduction to InVEST 10 InVEST User Guide, Release 2.5.6
TIER 0 Models TIER 1 Models TIER 2 Models TIER 3 Models Relative values Absolute values Absolute values Absolute values No valuation Valuation done through a suite Valuation done Valuation done through a of approaches through a suite of suite of approaches approaches Generally not time-specific, Annual average time step, no Daily to monthly Daily to monthly time or annual average temporal dynamics time step, some step, temporal dynamics temporal with feedbacks and dynamics thresholds Appropriate spatial extent Appropriate spatial extent Appropriate Appropriate spatial extent ranges from sub-watershed ranges from sub-watershed to spatial extent ranges from parcel to to global global ranges from global parcel to global Good for identifying key Good for strategic decisions Good for tactical More precise estimates of areas (relatively high risk or with absolute values, can be decisions with environmental service environmental service good for tactical decisions with absolute values delivery provision) calibration Some environmental service Some environmental service Some Sophisticated interactions interactions environmental environmental service service interactions with interactions feedbacks and thresholds
2.4 A work in progress
The development of InVEST is an ongoing effort of the Natural Capital Project. The models included in this release are at different stages of development and testing, however they are all sufficiently developed to be applied. To date, the terrestrial models have been applied in several sites and decision contexts, including to support: policy and conservation planning in the Willamette Basin USA, private landowners in Hawai’i USA, multi-stakeholder planning in Tanzania, permitting and licensing in Colombia, water fund design in Colombia and Ecuador, and priority setting for international aid in the Amazon Basin. We are currently applying the marine models in a number of places. Some examples include marine spatial planning in Canada and in Belize and climate adaptation planning in California and Texas. Updated and new models for additional environmental services will be released as they become available. InVEST is a freely available, open source product and as such the source code of each model can be inspected and modified by users. InVEST is subject to standard open source license and attribution requirements, which are described and must be agreed to in the installation process. A note on InVEST versioning: Integer changes will reflect major changes (like the addition of marine models war- ranted moving from 1.x to 2.0). An increment in the digit after the primary decimal indicates major new features (e.g the addition of a new model) or major revisions. We add a third decimal to reflect minor feature revisions or bug fixes with no new functionality. For example, InVEST 2.1.3 indicates the third iteration of the InVEST 2.1 models.
2.5 This guide
This guide will help you understand the basics of the InVEST models and start using them. The next chapter leads you through the installation process and provides general information about the tool and interface. The remaining chapters present the environmental service models. Each chapter: • briefly introduces a service and suggests the possible uses for InVEST results; • explains intuitively how the model works, including important simplifications, assumptions, and limitations;
2.4. A work in progress 11 InVEST User Guide, Release 2.5.6
• describes the data needed to run the model, which is crucial because the meaning and value of InVEST results depend on the input data; • provides step-by-step instructions for how to input data and interact with the tool; • offers guidance on interpreting InVEST results; • includes an appendix of information on relevant data sources and data preparation advice (this section is variable among chapters, and will improve over time from user input). This guide does not include detailed theoretical discussions of the scientific foundation of the models. For the terrestrial and freshwater models, these can be found in the new book Natural Capital: The Theory & Practice of Mapping Ecosystem Services (Oxford University Press).
2.5. This guide 12 CHAPTER THREE
GETTING STARTED
3.1 Getting started with InVEST
InVEST tools run as script tools in the ArcGIS ARCTOOLBOX environment. To run InVEST, you must have: • ArcGIS 9.3 (service pack 1 or 2) or ArcGIS 10 (service pack 1). • ArcINFO level license to run some of the models • Spatial Analyst extension installed & activated • The pollination model and all marine models require additional Python libraries available for download at www.naturalcapitalproject.org Running InVEST does not require Python programming, but it does require basic to intermediate skills in ArcGIS. A set of sample data is supplied with the models so you can become familiar with the models and how they work. To use InVEST for your context, however, you must compile the data described in the chapter(s) for the model(s) you wish to run and format them as indicated.
3.2 Installing the InVEST tool and data on your computer
The program InVEST-Setup.exe contains the InVEST toolbox, scripts, and training data, and is available for download at www.naturalcapitalproject.org. • Using Windows Explorer, take note of the folder structure and files extracted from InVEST-Setup.exe. Within the InVEST folder, you will see the toolbox InVEST220.tbx. The python scripts are in the folder \In- VEST220\python\. There is one script per model, and each ends with a *.py suffix. In addition, you will see folders for Base Data, Biodiversity, Hydropower, Carbon, and others. These folders contain sample data. The InVEST220.mxd file is an ArcMAP document with the InVEST toolbox and sample data pre-loaded.
3.3 Downloading and installing Python library extensions
InVEST users running the pollination model in ArcGIS 9.3 or any of the marine models in either ArcGIS 9.3 and 10 are required to download the Python extensions file found on the InVEST installer download page at www.naturalcapitalproject.org. The marine models include the “Marine Python Extension Check” tool found in the Marine InVEST toolset to determine which extensions are needed and if these extensions have been properly installed. Most Marine InVEST models require the following extensions to be installed:
13 InVEST User Guide, Release 2.5.6
1. Numeric Python (NumPy) is a powerful and flexible N-dimensional array container that provides the funda- mentals needed for scientific computing in Python. An older incompatable version of NumPy comes standard with the ArcGIS 9.3 and 10. While the “Marine Python Extension Check” tool will confirm that NumPy is al- ready installed on your machine, make certain to install the latest version of NumPy from the InVEST installer download page. 2. Scientific Library for Python (SciPy) is an Open Source library of scientific tools for Python. It calls on the NumPy library and gathers a variety of high-level science and engineering modules together as a single package. 3. Python for Windows (PythonWin) allows users to access data from Windows applications like Microsoft Excel. 4. Matplotlib is a Python 2D plotting library which produces publication quality figures. The Terrestrial InVEST Pollination model for ArcGIS 9.3 requires the installation of one extension: 1. Geospatial Data Abstraction Library (GDAL) is a translator library for raster geospatial data formats. See the Marine InVEST FAQ for help with installing these extensions.
3.4 Adding the InVEST toolbox to ArcMap
If you are working with sample data, you may wish to open InVEST220.mxd, which has the toolbox already loaded. Follow these steps if you will be working with your data. • START ArcMap. Save as a new mxd file. Ensure that ArcToolbox is open. If not, select the toolbox icon from the standard toolbar. • Right-click on an empty part of the ArcToolbox window and select ADD TOOLBOX. Or, right click on the top-most ArcToolbox text (see graphic below).
Figure 3.1: Adding the InVEST toolbox
• Navigate to the location of InVEST220.tbx, in the InVEST folder. Select the toolbox and click OPEN. Do not double click on the toolbox icon.
3.4. Adding the InVEST toolbox to ArcMap 14 InVEST User Guide, Release 2.5.6
Figure 3.2: Select InVEST Toolbox
• The INVEST toolbox should appear in ArcToolbox. Click on the plus sign to the left of InVEST to expand it. You will see scripts for each InVEST model.
3.5 Using sample data
The InVEST toolbox comes with sample data as a guide for formatting your data. For instance, in preparation for analysis of your data, you may wish to test the models by changing input values in the sample data to see how the output responds. Sample data are found in separate thematic folders in the InVEST folder. For example, the sample datasets for the Pollination model are found in \InVEST220\pollination\input, and those for the Carbon model in \In- VEST220\carbon\input. When opening the models, you’ll notice that default paths point to these sample datasets. You will also notice that the default workspace for each tool is the thematic folder with a name that matches the tool. Once you are working with your own data, you will need to create a workspace and input data folders that are structured like the sample data folders. You will also need to redirect the tool to access your data.
3.6 Formatting your data
Before running InVEST, it is necessary to format your data. Although subsequent chapters of this guide describe how to prepare input data for each model, there are several formatting guidelines common to all models: • Data file names should not have spaces (e.g., a raster file should be named ‘landuse’ rather than ‘land use’). • Raster dataset names cannot be longer than 13 characters and the first character cannot be a number. • Spatial data should be projected, and all input data for a given tool should be in the same projection. If your data is not projected InVEST will often give incorrect results. • Depending on the resolution (cell size) of your raster data, the model could take a long time to run. To make the tool run faster, enter a desired resolution that is larger than the original resolution. This will speed up the
3.5. Using sample data 15 InVEST User Guide, Release 2.5.6
Figure 3.3: InVEST Toolbox and tools displayed
execution, but will reduce the accuracy of your result. It is recommended to initially run models with large cell sizes to increase speed and reduce memory needs. Final results can be produced with finer resolution. • Results will be calculated on selections in tables and feature classes. If you are setting the model to read layers and tables from your ArcMap document rather than from the c-drive, make sure to clear any selections unless you wish to run your model on the selection. • Running the models with the input data files open in another program can cause errors. Ensure that the data files are not in use by another program to prevent data locking. • As the models are run, it may be necessary to change values in the input tables. This can happen within ArcMap or in an external program. Depending on the format of tables used (dbf or mdb is recommended) you will need an appropriate software program to edit tables. To edit tables within ArcMap, you need to start an edit session (from the editor toolbar) and select the workspace (folder or database) that contains your data. After editing you must save your changes and stop the edit session. • Some models require specific naming guidelines for data files (e.g., Biodiversity model) and field (column) names. Follow these carefully to ensure your dataset is valid. • Remember to use the sample datasets as a guide to format your data.
3.7 Running the models
You are ready to run an InVEST model when you have prepared your data according to the instructions in the relevant chapter and loaded the InVEST toolbox to your ARCMAP document. To begin:
3.7. Running the models 16 InVEST User Guide, Release 2.5.6
• Although not necessary, it’s often useful to add your input layers to your ARCMAP document to examine them. Use the ADD DATA button to add input data for a given module. • View the attribute table by right-clicking on the layer and selecting OPEN ATTRIBUTE TABLE. You can change the symbology of an input layer by right-clicking on the layer name in the TABLE OF CONTENTS and selecting PROPERTIES, then clicking on the SYMBOLOGY tab.
Note: Some of the models make changes to the data tables as they run. Such models will not run correctly if the tables are added to the map as the data will be locked.
• Double-click the model you wish to run (e.g., Carbon) and complete the required parameters in the dialogue box that appears.
• The Carbon dialog is shown below as an example. Fields for which the entered path leads to a non-existent file will be marked with a red “x” next to the space for that variable. You can run the model with sample data as shown by the default paths, or navigate the paths to your data. Instructions specific for each model are in subsequent chapters. • Note that each tool has a place to enter a suffix to the output filenames. Adding a unique suffix prevents overwriting files produced in previous iterations. When all required fields are filled in, click the OK button on the interface. • Processing time will vary depending on the script and the resolution and the extent of the datasets in the analysis. Every model will open a window showing the progress of the script. Be sure to scan the output window for useful messages. Normal progress notes will be printed in black font. Informative messages that may or may not require changes to the data will be indicated in green font. Messages in red font indicate problems that have caused the model not to run. Read the green and red messages carefully to be aware of potential data problems or to determine why the model did not produce an output. • The model creates two folders in the workspace you selected: ‘intermediate’ and ‘output.’ After your script completes successfully, you can view the results by adding them from the folders to your ArcMap document using the ADD DATA button. View the attribute table and change SYMBOLOGY, by right-clicking on the layer name in the TABLE OF CONTENTS and selecting PROPERTIES, then clicking on the SYMBOLOGY tab.
3.8 Support information
Several regular training workshops on InVEST may be offered annually, subject to funding and demand. Information on these trainings will be announced on the support page and can be found at the Natural Capital Project website. This site is also a good source of general information on InVEST and other activities of the Natural Capital Project.
3.8. Support information 17 InVEST User Guide, Release 2.5.6
Figure 3.4: Progress dialog
3.8. Support information 18 InVEST User Guide, Release 2.5.6
For other issues please contact the software team lead directly at [email protected].
3.9 Model run checklist
Use this checklist to ensure that the models run successfully. • ArcGIS Version: As stated above not all ArcGIS versions are supported. Most models are tested in ArcGIS 9.3 SP2 or ArcGIS 10 (for the selected models currently supported). It is advisable to upgrade to one of these versions. • Python Extensions: For all marine models, ensure that the latest Python library extensions have been installed: 1) NumPy, 2) SciPy, 3) PythonWin, and 4) Matplotlib. Additionally, Microsoft Excel is required to run the marine models. For ArcGIS 9.3 users, the pollination model requires installation of the GDAL library. • Spatial Analyst extension: Most of the models require ArcGIS spatial analyst extension. Ensure that this is installed. • Regional and Language options: Some language settings cause errors while running the models. For example settings which use coma (,) for decimals instead of period (.) cause errors in the models. To solve this change the regional settings to English. • Folder naming: ArcGIS is strict about folder naming. Avoid spaces and special characters in file and folder names.
3.10 Reporting errors
If you experience errors running the models please contact the software team lead directly at [email protected] and send the following information: • ArcGIS version and service pack number • InVEST model you’re having difficulty with • Explicit error message or behavior • If possible, a screenshot of the state of your InVEST toolset when you get the error.
3.11 Working with the DEM
For the hydrology tools Water Purification: Nutrient Retention and Avoided Reservoir Sedimentation, having a well- prepared digital elevation model (DEM) is critical. It must have no missing data or circular flow paths and should correctly represent the surface water flow patterns over the area of interest in order to get accurate results. Here are some tips for working with the DEM and creating a hydrologically-correct DEM. Included is information on using built-in ArcMap Spatial Analyst functions as well as ArcHydro (see resources below), an ArcMap data model that has a more complex and comprehensive set of tools for modeling surface water features. ArcSWAT, which is not covered here, could be a good option for delineating sub-watersheds. This is only intended to be a brief overview of the issues and methods involved in DEM preparation. For more detail, see the Resources section below. • Use the highest quality, finest resolution DEM that is appropriate for your application. This will reduce the chances of there being sinks and missing data, and will more accurately represent the terrain’s surface water flow, providing the amount of detail that is required for making informed decisions at your scale of interest.
3.9. Model run checklist 19 InVEST User Guide, Release 2.5.6
• The Hydro_layers directory When tools are run that use DEM-derived layers like slope and flow direction, the tool looks for a folder called ‘Hydro_layers’, located in the same folder as the DEM. If this folder does not exist, or any of the required derived layers within the folder don’t exist, the tool will generate them from the input DEM, otherwise it uses the layers that already exist. In general, this is convenient and efficient. However, if you decide to use a different DEM than the one that was used to generate the files in Hydro_layers, and the new DEM is located in the same folder as the old DEM, the tool will not realize that it is different, and will continue to use the old derived layers. So in this case it is necessary to delete the Hydro_layers folder before re-running the tool using the new DEM, so that the derived layers are regenerated. • Mosaic tiled DEM data If you have downloaded DEM data for your area that is in multiple, adjacent tiles, they will need to first be mosaicked together to create a single DEM file. In ArcToolbox, use Data Management -> Raster -> Mosaic to New Raster, entering all of the tiles into the Input Rasters list. Look closely at the output raster to make sure that the values are correct along the edges where the tiles were joined. If they are not, try different values for the Mosaic Method parameter to the Mosaic to New Raster tool. • Check for missing data After getting (and possibly mosaicking) the DEM, make sure that there is no missing data (or ‘holes’), repre- sented by NoData cells within the area of interest. If there are NoData cells, they must be assigned values. For small holes, one way to do this is to use the ArcGIS Focal Mean function within Raster Calculator (or Conditional -> CON). For example: con(isnull([theDEM]), focalmean([theDEM], rectangle,4,4), [theDEM])
Interpolation can also be used, and can work better for larger holes. Convert the DEM to points using Conversion Tools -> From Raster -> Raster to Point, interpolate using Spatial Analyst’s Interpolation tools, then use CON to assign interpolated values to the original DEM: con(isnull([theDEM]), [interpolated_grid], [theDEM])
Another possibility is assigning data from a different DEM, if surrounding values are a good match, again using CON: con(isnull([theDEM]), [different_DEM], [theDEM])
• Verify the stream network If the stream network generated from the DEM does not correctly match reality, ‘burning’ a correct stream network into the DEM might be necessary. Here are the basic steps for ArcMap: 1. Create the stream network from the DEM using the Hydrology -> Flow Accumulation tool and compare it to a known correct stream layer. If the generated stream network does not look correct, continue with the following steps. 2. If starting with a vector stream layer, convert it to a grid that has the same cell size and extent as the DEM. 3. Assign the stream grid a cell value of 1 where there are streams and 0 elsewhere. 4. Subtract a multiple of this stream grid from the DEM. If using ArcHydro, create the stream network from the DEM using Terrain Preprocessing -> Stream Definition and compare it to a known correct stream layer. If the generated stream network does not look correct, ‘burn’ the correct stream layer in using the Terrain Preprocessing -> DEM Manipulation -> DEM Reconditioning function. • Identify sinks in the DEM and fill them From the ESRI help on “How Sink works”: “A sink is a cell or set of spatially connected cells whose flow direction cannot be assigned one of the eight valid values in a flow direction raster. This can occur when all
3.11. Working with the DEM 20 InVEST User Guide, Release 2.5.6
neighboring cells are higher than the processing cell or when two cells flow into each other, creating a two-cell loop.” Sinks are usually caused by errors in the DEM, and they can produce an incorrect flow direction raster. Possible by-products of this are areas with circular flow direction (or a ‘loop’) or a discontinuous flow network. Filling the sinks assigns new values to the anomalous processing cells, such that they are better aligned with their neighbors. But this process may create new sinks, so an iterative process may be required. In ArcMap, first identify sinks using ArcMap’s Hydrology -> Sink tool. Fill the resulting sinks with Hydrology -> Fill. Do further iterations if there are still sinks that need to be filled. In ArcHydro, the corresponding tools are Terrain Preprocessing -> DEM Manipulation -> Sink Evaluation and Fill Sinks. • Flow direction loops If there’s a problem in the flow direction raster, such as a loop, the Water Purification and Sedimentation tools may go into an infinite loop and eventually time out, producing this error: “Error: Sub-watershed 1 is taking too long (45 minutes). This probably indicates that there’s a flow direction loop.” Diagnosing and repairing loops is difficult and is beyond the scope of our tools and built-in ArcMap functions. However, a very rough method of determining whether a loop is being encountered is provided in both of the scripts WP_2_Nutrient_Removal.py and Sediment_1_Soil_Loss.py. In each of these files, look for 3 separate commented-out sections of code beginning with ‘Flow direction loop debugging’. Uncomment the subsequent lines (containing references to ‘outfile’) as directed. The next time the tool is run, it will write information to the file:
This can become a very large file, as information is recorded on every cell in the watershed raster, as they are processed by moving along flow paths. Each line of the debug file has three values: the nutrient or sediment load originating on that cell, the flow direction and the fraction of nutrient or sediment retained by that land use class (as given in the input Biophysical table). With the debugging lines of code uncommented, run the tool. Then look at the end of the debug file - if a loop was encountered, multiple lines with a particular set of values will be repeated. These values can be used to help identify where the loops occur, by retaining the
3.11. Working with the DEM 21 InVEST User Guide, Release 2.5.6
• Creating sub-watersheds Sub-watersheds are now required for all of the InVEST hydrology models. For the Water Purification and Sediment models, each sub-watershed must be smaller than the equivalent of approximately 4000 x 4000 pixels, due to limitations with Python and the ArcMap memory model. To create sub-watersheds in ArcMap, use the Hydrology -> Watershed tool. In this case, the input point data will represent multiple points along the stream network within the main watershed, such that a sub-watershed will be generated for each. In ArcHydro, use the Watershed Processing -> Batch Subwatershed Delineation tool, with input point data representing multiple points along the stream network within the main watershed. A sub-watershed will be generated for each point. Again, after the sub-watersheds are generated, verify that they represent the catchments correctly. Ensure each sub-watershed is assigned a unique integer ID and that no duplicates are present.
3.12 Resources
ArcHydro: http://www.crwr.utexas.edu/giswr/hydro/ArcHOSS/Downloads/index.cfm ArcSWAT: http://swatmodel.tamu.edu/software/arcswat For more information on and an alternate method for creating hydrologically correct surfaces, see the ESRI help on “Hydrologically Correct Surfaces (Topo to Raster)”. For more information on sinks, see the ESRI help on “Creating a depressionless DEM”. Much more information and tips for all of these processes can be found by searching the ESRI support website.
3.12. Resources 22 Part II
Marine Models
23 CHAPTER FOUR
WAVE ENERGY MODEL
4.1 Summary
Decision-makers and the public are increasingly interested in converting wave energy into electricity, with the hope that ocean waves will be a source for clean, safe, reliable, and affordable energy. The goals of the InVEST wave energy model (WEM) are to map and value the energy provisioning service provided by ocean waves and to allow for the evaluation of trade-offs that might arise when siting wave energy conversion (WEC) facilities. The WEM assesses potential wave power and harvested wave energy based on wave conditions (e.g., significant wave height and peak wave period) and technology-specific information of WEC devices (e.g., performance table and maximum capacity). The model then evaluates the net present value (NPV) of building and operating a WEC facility over its life span using economic parameters (e.g., price of electricity, discount rate, as well as installation and maintenance costs). Obtaining accurate input data and parameters for the economic valuation portion of the model is a significant challenge because there have been no commercial-scale wave energy facilities implemented to date. We recommend using the NPV values of a wave energy facility computed with the default values be used to only to make relative comparisons between sites. The outputs of the WEM provide spatially explicit information, showing potential areas for siting WEC facilities with the greatest energy production and value. This site- and device-specific information for the WEC facilities can then be used to identify and quantify potential trade-offs that may arise when siting WEC facilities. Decision-makers and stakeholders can use the WEM to better understand where to install a WEC facility with greatest harvested wave energy and least effect on coastal and ocean ecosystems and other human uses. This is a “Tier 1” model.
4.2 Introduction
Wave energy has many characteristics important to the efficient generation of electricity and is considered a poten- tially significant contributor to the effort to meet growing human energy demands (Barstow et al. 2008). Among various renewable energy resources, wave energy has the greatest power density and provides relatively continuous and predictable power-significant advantages for electrical grid operation (Bedard et al. 2005). The cost of electricity generated by wave energy has decreased since the 1980s and is likely to decrease further as the technology devel- ops and the wave energy industry expands (Thorpe 1999). Considering the increasing cost of fossil fuel energy and concomitant interest in renewable energy sources, wave energy may be economically feasible in the near feature. As a consequence, decision-makers and the public are increasingly interested in converting wave energy into electricity with the hope that ocean waves will be a source for clean, safe, reliable, and affordable energy source without signifi- cant greenhouse gas emissions. With this increasing interest in wave energy as a renewable energy resource, there is a growing need for a framework to help decision-makers site wave energy facilities. The WEM we articulate here will provide planners with information that can be used to balance the harvesting of energy from waves with existing uses of marine and coastal ecosystems. Globally, exploitable wave energy resources are approximately equal to 20% of current world electricity consumption, but their potential varies considerably by location (Cornett 2008). In addition, in areas close to the shore, “hot spots”
24 InVEST User Guide, Release 2.5.6
(characterized by condensed wave energy) provide the highest potential for wave energy harvesting (Cornett and Zhang 2008, Iglesias and Carballo 2010). Therefore, indentifying wave-power-rich areas is the first step in siting a wave energy conversion (WEC) facility. A variety of technologies for WEC devices have been proposed to capture the energy from waves, and the particular characteristics of these devices play a critical role in quantifying the amount of energy that can be captured. Therefore, the choice of WEC device is also an essential component in efficiently harvesting wave energy under different wave conditions by location. The economic valuation of a wave energy facility can be used to compare the net benefits across sites and device- specific technologies. As with most renewable energy projects, many different factors can be included in the economic valuation. These include: the value of energy provided to the electricity grid, reduction in pollution associated with wave energy projects as compared to traditional sources, costs to those who lose access to coastal and marine locations, and environmental costs associated with the construction and operation of these facilities. In practice, including all the relevant benefits and costs, particularly those related to environmental benefits and costs, can be difficult to measure and include in a formal cost-benefit analysis. Rather than ignoring these potential impacts, we have taken a simple approach to incorporating some of this information in a simple framework that can be used in parallel to a formal cost-benefit analysis. While wave energy may provide clean and renewable energy without significant greenhouse gas emissions, wave energy projects may conflict with existing ocean uses or conservation strategies for protecting marine species and habitats. WEC facilities have the potential to impact fishing opportunities, pelagic and benthic habitat, recreational activities, aesthetic views, hydrodynamic and wave environments, navigation, and the bioaccumulation of toxic ma- terials (Boehlert et al. 2007, Nelson et al. 2008, Thorpe 1999). The severity of these potential impacts is likely to be site specific. Also, given limited experience with wave energy projects to date, there is little empirical evidence describing impacts. Therefore, identifying and evaluating the potential trade-offs associated with siting WEC facili- ties is an essential component of marine spatial planning and other forms of decision-making in marine and coastal environments. The WEM presented here assesses: 1) potential wave power, 2) harvested wave energy, and 3) the net present value of a WEC facility. The outputs of the WEM provide spatially explicit information, showing potential areas for siting WEC facilities with the greatest energy production and benefits. This site- and facility-specific information then can be used to evaluate how siting a WEC facility might influence and/or change existing coastal and marine uses. For example, the WEM allows users to explore potential trade-offs by mapping and quantifying spatial competition with existing ocean uses for commercial and recreational activities (e.g., fishing, navigation, whale watching, kayaking, etc.).
4.3 The model
The objective of the WEM is to help decision-makers and stakeholders inform marine spatial planning in the context of wave energy projects by exploring potential costs and benefits of siting wave energy facilities. The model can run using default input data sets that are globally and regionally available or with local input data.
4.3.1 How it works
Potential wave power resource assessment
Wave power per unit width of wave crest length transmitted by irregular waves can be approximated as ρ ∗ g P = H2C (T , h) (4.1) n 16 s g e
4.3. The model 25 InVEST User Guide, Release 2.5.6
-3 -2 where, Pn is wave power (kW/m), ρ is sea water density (1,028 kg m ), g is gravitational acceleration (9.8 m s ), Hs -1 is significant wave height (m), and Cg is wave group velocity (m s ) as a function of wave energy period, Te (sec), and water depth h (m) (Cornett 2008). Cg can be estimated as
2kh p g 1 + sinh(2kh) k tanh(kh) C = (4.2) g 2
where the wave number k is calculated using a dispersion relationship expressed as a function of wave frequency (w = 2π/Te) and water depth h:
w2 = gk ∗ tanh(kh) (4.3)
An iterative numerical solution scheme can be applied to solve Equation (4.3) with initial estimates of k = 2 p 2 w /(g · tanh(w · h/g)). The wave period of measured or modeled sea states are rarely expressed as Te, rather, they are often specified as peak wave period, Tp. Therefore, the peak energy period is estimated as Te = α · Tp. Where, α is constant determining the shape of a wave spectrum. We use α = 0.90 as a default value assuming standard JONSWAP spectrum, which works well when sea state is dominated by waves from a single source and the spectrum is unimodal (Cornett 2008). The same assumption was also applied to global wave power resource estimation (Cornett 2008) and wave power calculations in the west coast of Canada (Cornett and Zhang 2008, Dunnett and Wallace 2009). We prepared globally and regionally available input data layers for the potential wave power resources calculation. We used NOAA WAVEWATCH III (NWW3) model hindcast reanalysis results (version 2.22) to obtain wave charac- teristics defined by Hs and Tp. NWW3 spatial resolution ranges from 4 to 60 minutes depending on the global and regional grid systems. We used ETOPO1 to obtain the water depth (h), which provides 1 arc-minute global ocean bathymetry information (Amante and Eakins 2009). When using the default input data layers, model results provide the first approximation of potential wave power resources for any target area in the world. However, the spatial reso- lution of the model results may not be fine enough to assess wave power resources near coastal areas. So, this module will allow users to add their own wave input based on local studies (e.g., nearshore wave model results) in the next version.
Captured wave energy assessment
Captured wave energy can be estimated as a function of sea states and the wave energy absorption performance of a WEC device (Previsic 2004a, Previsic 2004b). A seastate is the general condition of the ocean surface and often characterized by two parameters, a significant wave height Hs and a peak period Tp. Long-term wave time-series data can be used to calculate the number of hours that each seastate occurs over a particular time period. We prepared globally and regionally available seastate tables using 3-hour interval NWW3 model results over a period of 5 years. Table 3.1 is an example of yearly occurrence of hours at each seastate bin in the west coast of Vancouver Island. In this example, a seastate with Hs = 2.5 m and Tp = 10.0 sec is most dominant, occurring 115 hours per year. The ability of a WEC device to harvest wave energy can be expressed by wave energy absorption performance that is available from WEC device manufacturers. We have conducted a literature review of WEC devices for which there is public information and prepared wave energy absorption performance tables for several WEC devices that have undergone full-scale testing and verification in the ocean. Currently, the InVEST WEM includes as default input parameters performance tables for: • PWP-Pelamis (Pelamis Wave Power Ltd 2010; Previsic 2004b) • Energetech-OWC (Previsic 2004a) • AquaBuOY (Dunnett and Wallace 2009) • WaveDragon (Dunnett and Wallace 2009)
4.3. The model 26 InVEST User Guide, Release 2.5.6
Figure 4.1: Occurrence of hours (hr/yr) in each seastate bin in the west coast of Vancouver Island.
Figure 4.2: Wave energy absorption performance (kW) in each seastate bin for Pelamis.
4.3. The model 27 InVEST User Guide, Release 2.5.6
By multiplying each cell in the annual occurrence of hours table by each corresponding cell of the wave energy absorption performance table, captured wave energy is calculated for each sea state bin. The annual captured wave energy (kWh/yr) per WEC device is calculated by summing up all the captured wave energy in each seastate bin. As technology progresses, the device-specific parameters will likely need to be updated and new devices added. Some WEC devices have an ability to optimize their performance in response to site-specific conditions (Previsic 2004b) and users may need to adjust the default parameters of a WEC device or apply their own machine-dependent values for more accurate assessments of harvested wave energy.
Net present value assessment
We used a cost-benefit analysis framework to evaluate the construction and operation of a wave energy facility. We combined the most relevant measures of benefits (Bt) and costs (Ct) to compute the NPV for a wave energy facility located at a specific location in marine space. The NPV of a particular wave energy facility is:
T X −t (Bt − Ct)(1 + i) (4.4) t=1 and is evaluated over a life span, T , of a WEC facility. To discount the value of future benefits and costs, we use a default discount rate, i, of 5 percent. Annual benefits are computed as the product of the price of electricity per kWh and annual captured wave energy in kWh 1. We assume no revenue in the initial year of the project. The annual costs can be broken down into initial installation costs and annual operating and maintenance costs. The initial costs of installing the wave energy devices include the following costs: 1) capital cost per installed kW, which is device dependent, 2) cost of mooring lines, 3) cost of underwater transmission cables, 4) cost of overland transmission cables 2. Because the costs of underwater and overland transmission cables depend on the distance of the facility to the nearest grid connection point, calculation of NPV allows users to evaluate the tradeoff between locating a facility in a particular location for its wave energy resources and the distance-dependent costs of installing the devices at that location. We provide default economic parameters tables for economic valuation of wave energy using three of the four machines described in the previous section: PWP-Pelamis, AquaBuOY and WaveDragon.
4.3.2 Limitations and simplifications
Some words of caution about limitations and simplifications of the model and guidance on model interpretation: 1. The quality of wave input data determines the accuracy of model results. So, a user needs to understand the quality of wave input data for proper interpretation of the WEM results. For example, the default wave input data are more appropriate for global and regional scale applications at 4 or 60 minutes spatial resolution. For a more detailed analysis of wave power in a region of interest, the user may want to provide wave model results obtained at a finer spatial resolution. 2. Captured wave energy indicates the yearly averaged energy absorbed per WEC device. For estimation of actual energy production from a WEC device, users may need to consider additional technology-specific information, such as device availability, power conversion efficiency, and directional factors. For some WEC devices, an increase in performance is possible without significant changes in the device structure and users may apply adjustment factors to the performance table. Please consult Previsic (2004a, 2004b) for further discussion about the estimation of actual wave energy production from a WEC facility.
1 Both the discount rate and the wholesale price of electricity are user-defined inputs. We provide a default value of 5% for the discount rate and .20 cents for the wholesale price of electricity. In many cases, fixed tariff or feed-in tariffs are being discussed to help promote development of renewable energy projects. 2 We do not consider the costs of additional land-based infrastructure that may be required to connect an offshore facility to the grid, nor do we consider the costs of permitting a wave energy project. Costs estimates for different wave energy conversion devices were derived from Dunnett and Wallace (2009) and converted to 2009 $USD.
4.3. The model 28 InVEST User Guide, Release 2.5.6
3. Because there have been no commercial-scale wave energy facilities implemented to date, obtaining accurate cost data is a challenge. We provide default values for several wave energy devices that are publicly available. Because these costs may be inaccurate and/or out of date, we recommend that NPV values of a wave energy facility computed with the default values be used to only to make relative comparisons between sites. These relative comparisons will highlight that potential wave power resources and distance to the grid will have a sig- nificant influence on the estimated project cost. The magnitude of the NPV computations should be interpreted with caution. 4. The cost estimates provided are scaled for a small to moderately sized wave farm 3. Larger farms would likely experience some cost savings from having to produce more machines, but might also require higher capacity and/or additional transmission cables. If you want to simulate the amount of energy harvested or the costs associated with a larger farm, you should carefully evaluate these factors. 5. The distance measure from a WEC facility to an underwater cable landing point is based on Euclidean metric and does not recognize any landmass within two target points. Users should be careful about distance estimation in regions with complex bathymetry.
4.4 Data needs
The model uses an interface to input all required and optional model data. Here we outline the options presented to the user via the interface as well as the maps and data tables used by the model. See the appendix for detailed information on data sources and pre-processing.
4.4.1 Required inputs
First we describe required inputs. The required inputs are the minimum data needed to run this model. The minimum input data allows the model to run globally without conducting economic analysis. 1. Workspace Location (required). Users are required to specify a workspace folder path. It is recommended that the user create a new folder for each run of the model. For example, by creating a folder called “runBC” within the “WaveEnergy” folder, the model will create “intermediate” and “output” folders within this “runBC” workspace. The “intermediate” folder will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “output” folder.: Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\WaveEnergy\runBC
2. Path to Folder with Wave Base Data (required). Users are required to specify the path on their system to the folder with input data for the Wave Energy model. When installing InVEST, about 1GB of global Wave Watch III wave data will be included.: Name: Path to a workspace folder. Avoid spaces. Sample path (default): \InVEST\WaveEnergy\Input\WaveData\
3. Analysis Area (required). This drop down box allows users to select the scale of their analysis and instructs the model as to the appropriate wave input data. Users will also have the option of selecting an area of interest (AOI, input #7, see optional inputs below). The AOI input serves to clip these larger areas in order to perform more detailed, local analysis. If an AOI is not specified, the model will conduct wave energy calculations for the entire analysis area. There are four preset areas: West Coast of North America and Hawaii, East Coast of North America and Puerto Rico, Global (Eastern Hemisphere), and Global (Western Hemisphere): File type: drop down options Sample (default): West Coast of North America and Hawaii
3 Wallace and Dunnett (2009) model 24 devices in their application.
4.4. Data needs 29 InVEST User Guide, Release 2.5.6
4. Machine Performance Table (required). This table indicates a machine’s “performance”, or its ability to capture wave energy given seastate conditions. The first row indicates wave period bins (Tp) in seconds while the first column indicates wave height bins (Hs) in meters. The remaining numbers in the table indicates captured wave energy for the given seastate condition defined by wave height (Hs) and period (Tp).: Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Excel 2007 or newer) Sample data set: \InVEST\WaveEnergy\Input\Machine_Pelamis.xls\Pelamis_performance$
5. Machine Parameters Table (required). This table indicates a machine’s maximum capacity and limits (wave height and period) to capturing wave energy given seastate conditions.: Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Excel 2007 or newer) Sample data set: \InVEST\WaveEnergy\Input\Machine_Pelamis.xls\Pelamis_parameter$
6. Global Digital Elevation Model (DEM) (required). A bathymetric raster layer is required to calculate ocean depths in meters. This information is incorporated into potential wave power calculation and the economic analysisvaluation to determine the cost to send mooring cables to the ocean floor before running them to landing points. If the user specifies a raster input that doesn’t cover the entire AOI, then wave output results outside this coverage will not include wave power calculations. To ensure the model runs properly, make sure this input covers the analysis area specified in input #2 and #7. The default bathymetry data, global_dem, provides 1 arc-minute global bathymetry data. If you are using wave input data coarser than 1arc1 arc-minute resolution, we recommend using the global demDEM data.:
4.4. Data needs 30 InVEST User Guide, Release 2.5.6
Name: File can be named anything, but no spaces in the name and less than 13 characters Format: GIS raster file (e.g., ESRI GRID or IMG) with depth information in meters Sample data set (default): \InVEST\Base_Data\Marine\DEMs\global_dem
4.4.2 Optional inputs
The next series of inputs are optional, but may be required depending on other decision inputs. 7. Area of Interest (AOI) (required for economic valuation). If you would like to further narrow your analysis area, you can create a polygon feature layer that defines your area of interest. It instructs the model where to clip the input data and defines the exact extent of analysis. This input is only required, however, when running the economic valuation. At the start, the model will check that the AOI is projected in meters and the datum is WGS84. If not, it will stop and provide feedback.: Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\WaveEnergy\Input\AOI_WCVI.shp
8. Compute Economic Valuation? By checking this box, users will instruct the model to run the economic valua- tion of the model. Currently, valuation is only permitted for runs where there is an AOI (input #7). Additionally, the following inputs (#9-11) must be also be specified in order to output economic analysis. 9. Economic Parameter Table (optional, but required for economic valuation). When running the economic analysis, the user must enter a table that includes the price of electricity, machine setup and cable costs, and other valuation parameters for net present value (NPV) calculations.: Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Excel 2007 or newer) Sample data set: \InVEST\WaveEnergy\Input\Machine_Pelamis.xls\Pelamis_econ$
10. Landing and Power Grid Connection Point Table (optional, but required for economic valuation). When running the economic analysis, you must provide an Excel spreadsheet that specifies locations where machine cables would reach land and eventually the energy grid. A point ID, latitude and longitude coordinates and the type of point are required. The model will use this input to create a point feature class and project it based on the projection of the AOI input #4.: Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Excel 2007 or newer) Sample data set: \InVEST\WaveEnergy\Input\LandGridPts_WCVI.xls\WCVI$
4.4. Data needs 31 InVEST User Guide, Release 2.5.6
When filling out the tables with your own data, make sure to: • Specify latitude and longitude in decimal degrees (as shown below) • Only include the words “LAND” or “GRID” in the “TYPE” column. Use the “TYPE” field to differentiate between the two landing types.
11. Number of Machine Units (optional, but required for economic valuation). When running the economic analysis, the user must enter an integer value for the number of devices per wave energy facility. This value is used for determining total energy generated during the life span (25 years) of a wave energy conversion facility. To determine a reasonable number of machines to enter, we recommend that the user divide the maximum capacity of the machine (see input #5) by the desired amount of energy captured. For example, if the user desires 21,000 kW of captured wave energy, then the wave energy farm would have 28 Pelamis (maximum capacity is 750kW), or 84 AquaBuoy (maximum capacity is 250kW), or 3 WaveDragon (maximum capacity is 7000kW).
4.5 Running the model
Note: The word ‘path’ means to navigate or drill down into a folder structure using the Open Folder dialog window that is used to select GIS layers or Excel worksheets for model input data or parameters.
4.5.1 Exploring the workspace and input folders
These folders will hold all input, intermediate and output data for the model. As with all folders for ArcGIS, these folder names must not contain any spaces or symbols. See the sample data for an example.
Exploring a project workspace and input data folder
The /InVEST/WaveEnergy folder holds the main working folder for the model and all other associated folders. Within the WaveEnergy folder there will be a subfolder named ‘Input’. This folder holds most of the GIS and tabular data needed to setup and run the model. The following image shows the sample folder structure and accompanying GIS data. We recommend using this folder structure as a guide to organize your workspaces and data. Refer to the screenshots below for examples of folder structure and data organization.
4.5. Running the model 32 InVEST User Guide, Release 2.5.6
4.5.2 Creating a run of the model
The following example describes how to set up the Wave Energy model using the sample data provided with the InVEST download. We expect users to have location-specific data to use in place of the sample data. These instructions provide only a guideline on how to specify to ArcGIS the various types of data needed and do not represent any site- specific model parameters. See the Data needs section for a more complete description of the data specified below. 1. Click the plus symbol next to the InVEST toolbox.
2. Expand the Marine toolset and click on the Wave Energy script to open the model.
4.5. Running the model 33 InVEST User Guide, Release 2.5.6
3. Specify the Workspace. Open the InVEST workspace. If you created your own workspace folder (Step 1), then select it here. Select the WaveEnergy folder and click to set the main model workspace. This is the folder in which you will find the intermediate and final outputs when the model is run. 4. Specify the Folder with Wave Base Data. The model requires the folder location of the wave data. Click and path to the InVEST/WaveEnergy/Input folder. Select the WaveData folder and click to set the wave data folder. 5. Specify the Analysis Area. You can run the model at one of two scales: Regional (West or East Coast of NA), or Global (Eastern or Western Hemispheres).
6. Specify the Area of Interest (AOI). The model does not require an AOI, unless the user chooses to run the economic valuation. However, the AOI does permit the user to perform more local analysis if the analysis area (specified above) is too large. This example refers to the AOI_WCVI.shp shapefile supplied in the sample data. You can create an AOI shapefile by following the Creating an AOI instructions in the FAQ. Click and path to the InVEST/WaveEnergy/Input data folder. If you created your own Input folder in Step 2, then select it here. Select the AOI_WCVI.shp shapefile and click to make the selection. 7. Specify the Machine Performance Table. The model requires an Excel table of machine performance characteris- tics. Click and path to the InVEST/WaveEnergy/Input data folder. Double left-click Machine_AquaBuOY.xls and select the worksheet AquaBuOY_performance$. Then click to make the selection.
4.5. Running the model 34 InVEST User Guide, Release 2.5.6
Note: ArcGIS and the model may not recognize the Excel sheet as valid data if it is added to the ArcMap Data View. It is best to add Excel data directly to the model using the Open and Add buttons and navigating to the data.
8. Specify the Machine Parameters Table. The model requires an Excel table of the physical specifications for a specific type of wave machine. Click and path to the InVEST/WaveEnergy/Input data folder. Double left-click Machine_AquaBuOY.xl*s and select *AquaBuOY_parameter$. Click to make the selection. 9. Specify the Digital Elevation Model. The digital elevation model provides the base data for the Wave Energy model. Click and path to the InVEST/BaseData/Marine/DEMs data folder. Select the global_dem raster, click to make the selection and add it to the Wave Energy model dialog window. 10. Specify the Economic Valuation (Optional). To conduct economic valuation of the wave energy conversion machines, click the checkbox. Economic analysis is only available if an AOI was specified. 11. Specify the Machine Economic Parameters Table (Optional). To conduct the economic analysis the model re- quires a table of economic valuation parameters. Click and path to the InVEST/WaveEnergy/Input data folder. Double left-click Machine_AquaBuOY.xls and select AquaBuOY_econ$. Make sure you select the work- sheet that corresponds to the correct wave machine specified in Steps 7 and 8. Click to make the selection. 12. Specify the Landing and Grid Points Table (Optional). To conduct the economic analysis the model requires an Excel table of machine locations. Click and path to the InVEST/WaveEnergy/Input data folder. Double left-click WCVI_LandGridPts.xls and select WCVI$. Click to make the selection. 13. Specify the Number of Machine Units (Optional). The model requires the number of machines to perform the economic valuation. Enter the number of machines as an integer by typing directly into the text box. 14. At this point the model dialog box is completed for a complete run of the Wave Energy model. Click to start the model run. The model will begin to run and will show a progress window with progress information about each step in the analysis. Once the model finishes, the progress window will show all the completed steps and the amount of time necessary for the model run.
4.5.3 Viewing output from the model
Upon successful completion of the model run, you will see new folders in your Workspace called “intermediate” and “Output”. The Output folder, in particular, may contain several types of spatial data, which are described in the Interpreting results section of this guide. You can view the output spatial data in ArcMap (from either the Intermediate or Output folders) using the “Add Data” button .
You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting “Properties”, and then “Symbology”. There are many options here to change the way the data appear in the map. You can also view the attribute data of output files by right clicking on a layer and selecting “Open Attribute Table”.
4.6 Interpreting results
4.6.1 Model outputs
The following is a short description of each of the outputs from the Wave Energy model. Each of these output files is automatically saved in the “Output” & “Intermediate” folders that are saved within the user-specified workspace directory:
4.6. Interpreting results 35 InVEST User Guide, Release 2.5.6
4.6. Interpreting results 36 InVEST User Guide, Release 2.5.6
Output folder
• Output\wp_kw & Output\wp_rc – These raster layers depict potential wave power in kW/m for the user-specified extent. The latter (“_rc”) is the former reclassified by quantiles (1 = < 20%, 2 = 20-40%, 3 = 40-60%, 4 = 60-80%, 5 = > 80%). – The potential wave power map indicates wave power resources based on wave conditions. These often provide the first cut in the siting process for a wave energy project. • Output\capwe_mwh & Output\capwe_rc – These raster layer depict captured wave energy in MWh/yr per WEC device for the user-specified extent. The latter (“_rc”) is the former reclassified by quantiles (1 = < 20%, 2 = 20-40%, 3 = 40-60%, 4 = 60-80%, 5 = > 80%). – The captured wave energy map provides useful information to compare the performance of different WEC devices as a function of site-specific wave conditions. • Output\npv_usd & Output\npv_rc – These raster layers depict net present value in thousands of $ over the 25 year life-span of a WEC facility for the user-specified extent. The latter (“_rc”) is positive values of the former reclassified by quantiles (1 = < 20%, 2 = 20-40%, 3 = 40-60%, 4 = 60-80%, 5 = > 80%). – The NPV map indicates the economic value of a WEC facility composed of multiple devices. A positive value indicates net benefit; a negative value indicates a net loss. Such information can be used to locate potential areas where a wave energy facility may be economically feasible. – These are only an output if you have chosen to run economic valuation. • Output\LandPts_prj.shp and GridPt_prj.shp – These feature layers contain information on underwater cable landing location and power grid connec- tion points, which have been projected based on the projection specified (input #12) and the coordinates specified in the Excel table for input #10. – The landing and grid connection points provide useful information for interpreting the NPV map. – It is only an output if the user chooses to run the economic valuation. • Parameters_[yr-mon-day-min-sec].txt
4.6. Interpreting results 37 InVEST User Guide, Release 2.5.6
– Each time the model is run a text file will appear in the workspace folder. The file will list the parameter values for that run and be named according to the date and time. – Parameter log information can be used to identify detailed configurations of each of scenario simulation.
Intermediate folder
• intermediate\WEM_InputOutput_Pts.shp – These point layers from the selected wave data grid are based on inputs #2-4. – They contain a variety of input and output information, including:
* I and J – index values for the wave input grid points * LONG and LAT – longitude and latitude of the grid points * HSAVG_M – wave height average [m] * TPAVG_S – wave period average [second] * DEPTH_M – depth [m] * WE_KWM – potential wave power [kW/m] * CAPWE_MWHY – captured wave energy [MWh/yr/WEC device] * W2L_MDIST – Euclidean distance to the nearest landing connection point [m] * LAND_ID – ID of the closest landing connection point that is closest * L2G_MDIST – Euclidean distance from LAND_ID to the nearest power grid connection point [m] * UNITS – number of WEC devices assumed to be at this WEC facility site * CAPWE_ALL – total captured wave energy for all machines at site [MWh/yr/WEC facility] * NPV_25Y – net present value of 25 year period [thousands of $] – The model outputs in raster format are interpolated results based on these point data. So, you can use this point information to explore the exact values of essential inputs and outputs at wave input data point locations. • intermediate\GridPt.txt and LandPts.txt + These text files log records of the grid and landing point coordinates specified in the Excel table for input #9 + This is only an intermediate output if you choose to run economic valuation.
4.7 Case example illustrating results
The following example illustrates the application of the wave energy model to the west coast of Vancouver Island (WCVI). The figures and maps are for example only, and are not necessarily an accurate depiction of WCVI. In this example, we use input data layers including: 1. Wave base data = West Coast of North America with 4-minute resolution 2. Area of Interest = AOI_WCVI.shp 3. WEC device = Pelamis 4. Digital Elevation Model = global_dem 5. Landing and Power Grid Connection Points = LandGridPts_WCVI.shp 6. Number of Machine Units = 28 7. Projection = WGS 1984 UTM Zone 10N.prj In order to generate a grid-scale power producing facility, it is necessary to capture a minimum of 10 kW/m of wave power (Spaulding and Grille 2010). Along the WCVI, this threshold is generally met, with the annual mean wave power >10 kW/m in most areas. Wave power gradually increases offshore. Approximately 20 kW/m wave power is
4.7. Case example illustrating results 38 InVEST User Guide, Release 2.5.6 available within 10 km of the shore, but the maximum wave power, 30-40 kW/m, is available 20-60 km offshore where depth is > 150 m.
Figure 4.3: Wave power potential (kW/m) in the west coast of Vancouver Island.
Captured wave energy in this example is calculated based on Pelamis devices with 750 kW power rating. The overall patterns of the captured wave energy are similar to those of potential wave power. A Pelamis device located at the 50-70 m depth contour produces approximately 2,000-2,300 MWh/yr of energy. Assuming 15 MWh/yr energy use per household in the WCVI (Germain 2003), each Pelamis unit produces enough energy to support 133-153 households.
Figure 4.4: Captured wave energy (MWh/yr) using a Pelamis device with a 750 kW power rating.
For the economic valuation of harvested wave energy, we calculate and map NPV over the 25-yr life-span of a WEC facility. For this example model run, each of the WEC facilities is composed of 28 Pelamis devices. We used an estimate of $100,000 for the underwater cable cost and 20 cents/kW for the price of electricity. Positive NPV occurs from 5-10 km offshore from the shoreline. It increases offshore and the highest NPV (the top 20% of all calculated NPV values ($4668k - $7307k)) occurs between 25-90 km from the shore.
4.7. Case example illustrating results 39 InVEST User Guide, Release 2.5.6
Figure 4.5: Net present value (thousand $) over a 25-yr life-span, using $100,000 per km for the cost of underwater transmission cables. Two underwater cable landing points are located in Tofino and Ucluelet (×) and the power grid connection point is located in Ucluelet (o). Each of the WEC facilities is composed of 28 Pelamis devices and the price of electricity is set at 20 cents per kW.
Because there have been no commercial-scale wave energy facilities implemented to date, large uncertainties exist in the economic parameters. In particular, the cost of underwater transmission cables is highly uncertain, ranging from $100,000 to $1,000,000 per km. The NPV uses a lower bound of $100,000 per km for the cable cost. When we use a median cost of underwater transmission cables ($500,000 per km), the area with a positive NPV is significantly reduced. In this example, positive NPV only occurs within a 50 km radius around the two underwater cable landing points in Tofino and Ucluelet. The upper 20% NPV exists between 10-40 km distances from the two landing points. When the upper bound ($1,000,000 per km) of transmission cable costs is used, no positive NPV exist in the WCVI. Consid- ering uncertainties in economic parameters, users should be cautious in interpreting the magnitude of the NPV. We recommend that the NPV of a wave energy facility computed with the default values be used only to make relative comparisons between sites.
Figure 4.6: Net present value (thousand $) over a 25-yr life-span, using $500,000 per km for the cost of underwater transmission cables. Two underwater cable landing points are located in Tofino and Ucluelet (×) and power grid connection point is located in Ucluelet (o). Each of the WEC facilities is composed of 28 Pelamis devices. The price of electricity is set at 20 cents per kW.
4.7. Case example illustrating results 40 InVEST User Guide, Release 2.5.6
4.8 Appendix A
4.8.1 Data sources
This is a rough compilation of data sources and suggestions for finding, compiling, and formatting data. This section should be used for ideas and suggestions only. We will continue to update this section as we learn about new data sources and methods.
• Wave data: significant wave height (Hs) and peak wave period (Tp) – Global ocean wave buoy data are available from NOAA’s National Data Buoy Center (http://www.ndbc.noaa.gov/). Although ocean wave buoy provides the most accurate wave time series data, their spatial resolution is very coarse and it may not be appropriate for local scale analysis. – NOAA’s National Weather Service provides WAVEWATCH III model hindcast reanalysis results (http://polar.ncep.noaa.gov/waves/index2.shtml). The spatial resolution of the model results ranges from 4 to 60 minutes depending on the global and regional grid systems. The model outputs have been saved at 3-hour interval from 1999 to the present. The model results have been validated with ocean buoy data at many locations and provide good quality wave information. • Water depth – NOAA’s National Geophysical Data Center (NGDC) provides global bathymetry data with various spatial resolutions at http://www.ngdc.noaa.gov/mgg/bathymetry/relief.html. – ETOPO1 is a 1 arc-minute global relief model of Earth’s surface that integrates land topography and ocean bathymetry. It was built from numerous global and regional data sets, and is available in “Ice Surface” (top of Antarctic and Greenland ice sheets) and “Bedrock” (base of the ice sheets) versions. NGDC also provides regional and other global bathymetry datasets. • Wave energy absorption performance – EPRI wave energy conversion project provides a review of several WEC devices: http://oceanenergy.epri.com/waveenergy.html – Recent updates on technology may be available from the WEC device manufactures.
* PWP-Pelamis: http://www.pelamiswave.com/ * AquaBuOY: http://www.finavera.com/ * WaveDragon: http://www.wavedragon.net/ * DEXAWAVE: http://www.dexawave.com/
4.9 Wave Energy 3.0 Beta
We are working on the next generation platform of InVEST and deploying parts of it as prototype InVEST models. Wave Energy has a 3.0 prototype which can be found in the InVEST 3.0 Beta toolbox inside the InVEST +VERSION+ toolbox. Currently it is only supported in ArcGIS 10. New features to the 3.0 version include: • Paramters from previous runs are automatically loaded into the user interface. • Runtime of the model has been improved. • The projection input is no longer required and we calculate distances from the projection given by the area of interest polygon layer. • All table inputs are now given in CSV format rather than Excel files.
4.8. Appendix A 41 InVEST User Guide, Release 2.5.6
Please send feedback or bug reports to [email protected].
4.10 References
Amante, C., and B. W. Eakins. 2009. ETOPO1 1 Arc-minute global relief model: procedures, data sources and analysis, p. 19. NOAA Technical Memorandum NESDIS NGDC-24. Barstow, S., G. Mørk, D. Mollison, and J. Cruz. 2008. The wave energy resource, p. 94-131. In J. Cruz [ed.], Ocean Wave Energy: current status and future prepectives. Springer. Bedard, R., G. Hagerman, M. Previsic, O. Siddiqui, R. Thresher, and B. Ram. 2005. Offshore wave power feasibility demonstration project: final summary report, p. 34. Electric Power Research Institute Inc. Boehlert, G. W., G. R. Mcmurray, and C. E. Tortorici. 2007. Ecological effects of wave energy development in the Pacific Nothwest, p. 174. U.S. Dept. Commerce, NOAA Tech. Memo. Cornett, A., and J. Zhang. 2008. Nearshore wave energy resources, Western Vancouver Island, B.C., p. 68. Canadian Hydraulics Centre. Cornett, A. M. 2008. A global wave energy resource assessment. Proc. ISOPE 2008. Dunnett, D., and J. S. Wallace. 2009. Electricity generation from wave power in Canada. Renewable Energy 34: 179-195. Germain, L. A. S. 2003. A case study of wave power integration into the Ucluelet area electrical grid. Master Thesis. University of Victoria. Iglesias, G., and R. Carballo. 2010. Wave energy and nearshore hot spots: the case of the SE Bay of Biscay. Renewable Energy 35: 2490-2500. Nelson, P. A. and others 2008. Developing wave energy in coastal California: potential socio-economic and environ- mental effects, p. 182. California Energy Commission, PIER Energy-Related Environmental Research Program, and California Ocean Protection Council. Pelamis Wave Power Ltd. 2010. Pelamis Wave Power. http://www.pelamiswave.com/. Previsic, M. 2004a. System level design, performance and costs - San Francisco California Energetech offshore wave power plant. EPRI. —. 2004b. System level design, performance and costs for San Francisco California Pelamis offshore wave power plant, p. 73. EPRI. Spaulding, M. L., and A. Grilli. 2010. Application of technology development index and principal component analysis and cluster methods to ocean renewable energy facility siting. Marine Technology Society Journal 44: 8-23. Thorpe, T. W. 1999. A brief review of wave energy, p. 186. The UK department of trade and industry. Wilson, J. H., and A. Beyene. 2007. California wave energy resource evaluation. Journal of coastal research 23: 679-690.
4.10. References 42 CHAPTER FIVE
WIND ENERGY MODEL
5.1 Summary
Offshore wind energy is gaining interest worldwide, with 5,400 megawatts (MW) installed as of January 2013 and a growth rate around 25% per year (GWEC, 2013). Consistently higher offshore winds and proximity to coastal load centers serve as two of the major reasons wind energy developers are looking offshore. The goal of the InVEST offshore wind energy model is to provide spatial maps of energy resource availability, energy generation potential, and (optionally) energy generation value to allow users to evaluate siting decisions, use tradeoffs, and an array of other marine spatial planning questions. The model was developed to allow maximum flexibility for the user, in that it can be run with default data and parameters, but it can just as easily be updated with new turbine and foundation information, grid connection information, and parameter values that fit the user’s context. Model outputs include wind power potential, energy generation, offset carbon emissions, net present value, and levelized cost of energy, all given at the farm level.
5.2 Introduction
This wind energy model provides an easily replicable interface to assess the viability of wind energy in your region under different farm design scenarios. The outputs are raster maps, whose point values represent the aggregate value of a farm centered at that point. This allows for detailed analysis of siting choices at a fine scale, though it comes at the cost of assuming that conditions are sufficiently symmetric around the center point so that the center point represents the median conditions of all turbines in the farm. Since the user can select the number of turbines for the farm, and the raster maps do not give an indication of farm size, the model also outputs a representative polyline polygon at a randomly selected wind data point that indicates the size of the farm. To run the model, you are asked to supply information into the graphical user interface. This includes information about the type of turbine, number of turbines, the area of interest, etc. To minimize the set of required inputs, the model includes default data in .csv tables on two common offshore wind turbines: 3.6 MW and 5.0 MW. Also included is a table of less commonly changed default values used to parameterize various parts of the model, called the “Global Wind Energy Parameters” file. These .csv files are required inputs, and may be modified if alternate values are desired by directly editing the files using a text editor or Microsoft Excel. When modifying these files, it is recommended that the user make a copy of the default .csv file so as not to lose the original default values.
43 InVEST User Guide, Release 2.5.6
5.3 The model
5.3.1 Wind Energy Potential
The wind energy model estimates wind power density (wind power potential) to identify offshore areas with high energy potential. The wind power density PD(W m−2) at a certain location can be approximated as a function of wind statistics (Elliott et al., 1986)
c 1 X 3 ρ f(Vj)V (5.1) 2 j j=1
−3 where, ρ is mean air density (kg m ), j is the index of wind speed class, c is the number of wind speed classes, Vj −1 is wind speed of the jth class (ms ), and f(Vj) is probability density function of Vj. Two probability distributions are commonly used in wind data analysis: 1) the Rayleigh and 2) the Weibull distributions (Manwell et al. 2009). The Weibull distribution can better represent a wider variety of wind regimes (Celik 2003; Manwell et al. 2009), and is given as
k−1 V k k Vj − j f(V ) = e λ (5.2) j λ λ
where, k and λ are the shape and scale factors, respectively. The shape factor, k, determines the shape of the Weibull probability density function (Figure 5.1). The probability density function shows a sharper peak as k in- creases,indicating that there are consistent wind speeds around the mean wind speed. On the other hand, the function becomes smoother as k decreases, indicating more variation in wind speed and more frequent low and high wind speeds. We used a MATLAB function, wblfit, to estimate k and λ, which returns the maximum likelihood estimates of the parameters of the Weibull distribution given the values in the wind time series data. For more details of wblfit function, please consult http://www.mathworks.co.kr/kr/help/stats/wblfit.html. Wind power density is calculated at the hub height Z (m) of a wind turbine (Figure 5.1), which means all variables in (5.1) and (5.2) need to be converted into the appropriate value at hub height. Mean air density ρ was estimated as ρ = 1.225 − (1.194 · 10−4)Z, which approximates the U.S. Standard Atmosphere profile for air density (National Oceanic and Atmospheric Administration, 1976). We applied the wind profile power law to estimate wind speed (V ) at hub height Z (Elliott et al., 1986).
V Z α = (5.3) Vr Zr
−1 −1 where V is wind speed (ms ) at the hub height Z (m) of a wind turbine, and Vris wind speed (ms ) at the reference height Zr (m) where wind data are obtained. α is power law exponent, which is an empirically derived coefficient and varies with the stability of the atmosphere. For neutral stability condition, α is approximately 1/7 (0.143) for land surfaces, which is widely applicable to adjust wind speed on land (Elliott et al., 1986). The power law exponent has different value on ocean surfaces. Hsu et al (1994) found that α = 0.11 ± 0.03 for ocean surface under near-neutral atmospheric stability conditions. The wind energy model uses α = 0.11 as a default value to adjust wind speed on the ocean surface. The wind profile of the atmospheric boundary layer can be approximated more accurately using the log wind profile equation that accounts for surface roughness and atmospheric stability (Manwell et al. 2009). Wind power density (PD) outputs provide suitability information for a wind energy development project in terms of wind resource. Pacific Northwest Laboratories categorized wind power density and wind speed into seven classes based on United States wind atlas (Figure 5.3) (Manwell et al. 2009). Areas designated as class 4 or greater are considered to be suitable for most wind energy development. Class 3 areas are suitable for wind energy development if large turbines are used. Class 1 and 2 are rarely considered as suitable areas for wind energy development in terms
5.3. The model 44 InVEST User Guide, Release 2.5.6
Figure 5.1: Example of Weibull probability density function with various shape factors (k), where mean wind velocity = 6ms−1 (Manwell et al., 2009).
Figure 5.2: A schematic diagram of a wind turbine (http://www.daviddarling.info/encyclopedia/H/AE_hub_height.html)
5.3. The model 45 InVEST User Guide, Release 2.5.6 of energy potential. Wind resources vary considerably over space and a more detailed categorization of wind power density for five topographical conditions was developed in Europe, which includes sheltered terrain, open plain, sea coast, open sea, hills and ridges (Figure 5.3) (Manwell et al. 2009). The wind resource classification for sea coast and open sea may provide better information on the suitability of offshore wind energy projects.
Figure 5.3: Wind power density (PD) and wind speed classes based on European wind atlas (Modified from Table 2.6 in Manwell et al. 2009).
5.3.2 Energy Generation
The amount of energy harvestable from a wind turbine in a particular location depends on the characteristics of the wind turbine as well as wind conditions (Pallabazzer 2003; Jafarian & Ranjbar 2010). The wind energy model quantifies the harvestable energy based on the output power curve of a wind turbine and wind speed statistics. Figure 5.4 shows an output power curve of a wind turbine (pitch control type). The wind turbine starts to generate power at the cut-in wind speed (Vcin). The output power increases up to the rated power (Prate) as wind speed increases to the rated wind speed (Vrate). The wind turbine keeps producing the maximum power (i.e., Prate) until wind speed reaches the cut-out wind speed (Vcout). If wind speed increases beyond the cut-out wind speed, the wind turbine stops generating power for safety purposes. Currently, more than 74 offshore wind farms are operating globally and technology specific information of the wind turbine at each wind farm are available at LORC Knowledge (2012).
Figure 5.4: Output power (P) curve of a wind turbine (pitch control type) as a function of wind speed (V) (Modified from Fig.1 in Pallabazzer 2003)
5.3. The model 46 InVEST User Guide, Release 2.5.6
To provide flexibility for a variety of different turbine types without requiring the user to manually enter in a power curve, we estimate the output power P (kW) of a wind turbine using a polynomial modeling approach (Jafarian & Ranjbar 2010): 0 V < Vcin or V > Vcout P (V ) = Prate Vrate < V < Vcout (5.4) m m m m (V − Vin )/(Vrate − Vin ) Vcin ≤ V ≤ Vrate where, m is an exponent of the output power curve (usually 1 or 2). Using this approach, the energy output, O (MWh), generated by a wind turbine can be calculated using ! Z Vr ate m m Z Vcout ρ V − Vcin O = nday · Prate m m f(V )dV + f(V )dV (1 − lossrate) (5.5) ρ0 Vcin Vr − Vcin Vrate
where, nday is the number of days for energy output (e.g. nday = 365 days for annual energy output), ρ0 is air density of standard atmosphere (e.g. 1.225kgm−3 for U.S. standard atmosphere air density at sea level), and lossrate is a decimal value which represents energy losses due to a combination of downtime, power conversion efficiency, and electrical grid losses (default value is .05). All of these parameters are included in the global parameters .csv file and may be changed by the user from their defaults. Total farm energy output is equal to the individual turbine output multiplied by the number of turbines, n, E = nO (5.6)
The InVEST software comes with default technical and financial information about two common turbine sizes, the 3.6 MW and 5.0 MW turbines. The information for each turbine is given in .csv files in the Input directory and is a required input into the model. The user can use the default data, edit a file, or create a new file to assess different turbine sizes or update specific characteristics. The files must retain the same format; only parameter values may safely be modified. It is recommended to save edits as new .csv files rather than overwriting the default data.
5.3.3 Offset Carbon
Since wind turbines create no greenhouse gasses when generating energy, the user may be interested in assessing the amount of carbon dioxide emissions avoided by building a wind farm versus a conventional energy generation plant. To translate carbon-free wind power to a representative amount of annual avoided CO2 emissions, we use the −4 following default conversion factor: 6.8956 · 10 metric tons CO2/kW h This is obtained from the EPA (http://www.epa.gov/cleanenergy/energy-resources/refs.html) and is based on 2007 data. See their website for limitations of this approach. The parameter is representative of the carbon emitted by the energy portfolio of the United States and may not be appropriate for your context. This value is changeable in the global parameters .csv file.
5.3.4 Value of Power
The value of wind power is measured as the discounted net revenue from power generation that would ac- crue to a wind power developer/operator over the expected lifetime of a wind farm. The Net Present Value (http://en.wikipedia.org/wiki/Net_present_value) (NPV) of energy for a given wind farm is:
T X −t NPV = (Rt − Ct)(1 + i) (5.7) t=1
5.3. The model 47 InVEST User Guide, Release 2.5.6
Where Rt is the gross revenue collected in year t, and Ct are the aggregate costs in year t. math:T represents the expected lifetime of the facility, and i represents the discount rate (http://en.wikipedia.org/wiki/Discount_rate) or weighted average cost of capital (WACC, http://en.wikipedia.org/wiki/Weighted_average_cost_of_capital). Both T and i can be changed by the user; T can be found in the global parameters .csv file and i is entered in the valuation section of the user interface. For projects that are financed by both debt and equity and where there is a significant amount of risk associated with establishing and maintaining the projected stream of revenues, WACC is a more appro- priate method for establishing the time value of money. As this parameter enters into the calculation in the same way as a discount rate would, if you prefer you can input an appropriate discount rate and interpret the results accordingly. We do not supply a default value, but Levitt et al. (2011) suggest a WACC value of .116 based on a comprehensive analysis of industry specific discount rates and different debt/equity structures in Europe and the U.S. This is higher than discount rates typically used elsewhere, such as in standard cost benefit analysis, so you may find your application justifies a different rate. Annual gross revenue is calculated by multiplying the price per kWh, s, by the amount of kWh supplied to the grid by a wind farm, Et, thus Rt = sEt. It is assumed that energy is not collected in the first year during the construction phase. Costs can be separated into one-time capital costs and ongoing operations and management costs. During the construc- tion phase, expenditures are made on turbines, foundations, electrical transmission equipment, and other miscellaneous costs associated with development, procurement, and engineering. At the end of the farms usable lifetime, the firm must remove their equipment. The default information supplied is based on an extensive review of peer-reviewed publications, industry reports, and press releases. This information is summarized below.
Turbines
Turbines and foundations are modeled with unit costs. We have supplied cost data on 3.6 MW and 5.0 MW class turbines as well as monopile and jacketed foundations, though you may enter your own turbine- or foundation-specific information. Note all default costs below are given in 2012 US dollars. Assuming one foundation per turbine, the total cost of turbines and foundations is simply the number of wind turbines multiplied by the unit cost. Table 1 gives a summary of existing turbine costs. Farm Location Capacity # of Turbines Total MW Unit Cost ($mil) Riffgat UK 3.6 30 108 6.3 Sheringham Shoal UK 3.6 88 317 5.65 Greater Gabbard UK 3.6 140 504 6.03 Butendiek Germany 3.6 80 288 5.54 London Array UK 3.6 175 630 6.29 Amrumbank Germany 3.6 80 288 6.41 Global Tech 1 Germany 5 80 400 10.4 Borkum 2 Germany 5 40 200 10.6 Table 1: Turbine costs.
Foundations
This model can flexibly include valuation for both foundation-based and floating turbine designs. This is accomplished by letting the user enter the appropriate unit cost information for their farm design. Outputs are constrained by user-editable depth and distance parameters, so it is important to adjust these to reflect the appropriate technological constraints of your design choice. Foundation-based turbines have conventionally been limited to a depth of around 60 meters. Foundation cost information is relatively difficult to come by. Monopile foundations are the most common foundation type and are typically mated to 3.6 MW turbines. Ramboll, a major foundation manufacturer, estimates that monopile foundations with a 3.6 MW turbine are $2 million per foundation. Monopile costs at Burbo and Rhyl Flats in the UK were given in press releases as $1.9 million $2.2 million respectively. Jacketed foundations are more robust than
5.3. The model 48 InVEST User Guide, Release 2.5.6 monopile foundations and are typically used with 5.0 MW turbines and/or in deep water. Two press releases for Nordsee Ost (Germany) and Ormonde (UK) put the unit costs for this type of foundation at $2.74 million and $2.43 million respectively. A 2012 release by the European Energy Programme for Recovery put the cost of deepwater (40 meters) gravity foundations at Global Tech 1 (Germany) as $6.65 million per foundation. All foundations should feature an increasing cost with depth as material costs will necessarily be higher; however, this is not captured in this model currently due to the paucity of project cost data to estimate such a relationship.
Electricity Transmission
Electricity transmission equipment is much harder to model at the component level because the optimal transmission system design varies considerably with local conditions and wind farm design. Depending on the size of the farm and its distance from shore, offshore platforms with voltage transformers, converters, and switchgear may be needed. Additionally, there is a critical point where a wind farm’s distance from the grid requires a switch from alternating current (AC) power to direct current (DC) power to overcome line losses which reduce the amount of energy delivered. Given design variation across different contexts, we utilized a top-down modeling approach for transmission costs to allow the model to be used broadly without the need for exhaustive system modeling and unit cost information. We collected information about electricity transmission costs (including installation) from 20 wind farms and used it to estimate a relationship between total costs and farm characteristics. This data was collected from the U.K. Ofgem tender process (http://www.ofgem.gov.uk/Networks/offtrans/Pages/Offshoretransmission.aspx) and is shown in Table 2. Farm Cost (2012 MW Depth DC Land Cable Sea Cable Tot Cable $Million) (m) (km) (km) (km) Barrow 52.73 90 14 0 3 27 30 Robin Rigg 93.25 180 6 0 1.8 12.5 14.3 Gunfleet Sands 1 75.8 173 6.5 0 3.8 9.3 13.1 & 2 Sheringham 285.08 315 18.5 0 21.5 22.4 43.9 Shoal Ormonde 158.27 150 19 0 2.8 43 45.8 Greater Gabbard 495.64 504 20.5 0 0.6 45.5 46.1 Thanet 255.34 300 18.5 0 2.4 26.3 28.7 Walney 1 173.56 183 21 0 2.7 45.3 48 Walney 2 164.38 183 27 0 5 43.7 48.7 Gwynt y Mor 449.38 576 27.5 0 11 41.4 52.4 Lincs 456.44 250 9.5 0 12 48 60 London Array 699.28 630 11.5 0 0.8 54 54.8 Phase 1 Nordergrunde 89.82 111 7 0 4 28 32 Dolwin 1 1221.81 800 30.5 1 90 75 165 Dolwin 2 1021.5 900 29 1 90 45 135 Helwin 2 817.51 690 28.5 1 45 85 130 Sylwin 1 1393.38 864 26 1 45 160 205 Helwin 1 718.71 576 23 1 45 85 130 Borwin 2 718.71 800 40 1 75 120 195 Borwin 1 598.83 400 40 1 75 125 200 Table 2: Offshore energy transmission infrastructure. Using an ordinary least squares regression, we estimated the following equation that relates total transmission costs to farm capacity and total transmission cable distance:
T ransCost = β0MW + β1T otCable + (5.8)
5.3. The model 49 InVEST User Guide, Release 2.5.6
To capture the effect of transmission losses due to resistance, we estimated this separately for each current type (AC and DC). Since our data suggest a critical threshold of greater than 54.8km for AC transmission, we adopt 60km as the transition point. This is also consistent with published figures regarding the cost effectiveness of transitioning from AC to DC transmission (Carbon Trust, 2008; UMaine, 2011); see Table 3 Costs if ≤ 60km (AC) Costs if > 60km (DC) MW .81*** 1.09** (.15) (.37) Cables 1.36 .89 (1.19) (1.61) Adj R2 .937 .951 Table 3, AC DC transmission costs. *p<.10, **p<.05, ***p<.01 These results provide a predictive model of transmission costs as a function of current type, total farm capacity in MW, and the total length of transmission cable in km. To calculate the total length of transmission cable from any given offshore location, the model requires some information about the onshore grid. The user has two options: • Create a .csv file that includes latitude and longitude details for all grid connection points in the area of interest • Use a fixed parameter to model grid location The fixed parameter option specifies a mean distance inland along the entire coast that represents the expected distance that overland cables may have to travel to reach a grid connection. Since grid connection points for large farms are very opportunistic and represent a relatively small portion of capital costs, it is not unrealistic to model grid connection this way in the absence of a detailed grid connection scheme. The default parameter included, 5.5 km, is the mean overland cable distance from the UK from the transmission infrastructure table above. Above and beyond the cost of sending the energy to shore, wind farms also require cables which connect turbines to each other, called array cables. We estimated a simple linear relationship between array cables and the number of turbines based on the data given below: Farm Location # of Turbines km of cable Total Cost ($mil) Nordsee Ost Germany 48 63 14.3 Amrumbank Germany 80 86 27.1 Gwynt y Mor UK 160 148 34.7 Anholt Denmark 111 160 41.4 Baltic 2 Germany 80 80 32.5 Sheringham Shoal UK 88 88 17.7 Table 4. Array cabling The data above suggest that .91km of cable is required per turbine at a cost of $260,000 per km. This establishes a relationship of array cable to wind turbines which can retrieve the total cost of array cable based only on the number of turbines in the farm.
Other Costs
There are a variety of additional costs associated with the construction phase, such as those for development, engi- neering, procurement, and royalties. AWS Truewind (2010) estimate these costs to amount to 2% of total capital expenditures; Blanco (2009) indicates it could be as high as 8%. We adopt their method of using a ratio of capital costs for calculating these costs and use the mean value of 5% as the default . Installation of foundations, turbines, and transmission gear (cables and substations) comprises its own cost category. Kaiser and Snyder (2012) take a comprehensive view of installation costs and find that installation costs make up ap- proximately 20% of capital expenditures in European offshore wind farms. Accordingly, this model treats installation costs as a fixed percentage of total capital costs and uses the default value suggested by Kaiser and Snyder (2012).
5.3. The model 50 InVEST User Guide, Release 2.5.6
Decommissioning the facility at the end of its useful life (t = T ) enters into the model in a similar way as installation costs, in that it is a fixed fraction of capital expenditures. Snyder and Kaiser (2012) put this one-time cost at 2.6% to 3.7% of initial expenditures (net of scrap value) for the Cape Wind farm using a sophisticated decommissioning model. The default value used in this model is 3.7%. Most of the costs of an offshore wind energy farm are related to the initial capital costs; however, there are ongoing costs related to maintenance and operations (O&M) as well. Boccard (2010) uses a methodology consistent with the rest of our modeling by calculating annual O&M cost as a % of original capital costs, and puts the costs somewhere between 3 and 3.5. The default value used in this model is 3.5%, and can be changed along with all the other costs in this section by editing the global parameters .csv file.
Energy Prices
This model is currently designed to accept a fixed unit price for a kilowatt hour (kWh) of energy over the lifetime of the wind farm. In some locations, wind farm operators receive a subsidized rate know as a feed-in tariff which guarantees them a set price for their energy over some time horizon. In other locations, wind farm operators must negotiate with energy providers and public utility commissions to secure a power purchase agreement. These are contracts that specify a unit price for energy delivered. We do not supply a default unit price for energy as energy prices fluctuate widely over space and government policies may significantly influences prices in different countries. Therefore it is worthwhile for the user fully investigate what price makes sense for their context.
5.3.5 Levelized Cost of Energy
The levelized cost of energy (http://en.wikipedia.org/wiki/Cost_of_electricity_by_source) (LCOE) is the unit price that would need to be received for energy that would set the present value of the project equal to zero. As such, it gives the lowest price/kWh that a wind farm developer could receive before they considered a project not worthwhile. The output given by the model is in terms of $/kWh and is calculated as:
PT O&M·CAP EX D·CAP EX t=1 t + T + CAP EX LCOE = (1+i) (1+i) PT Et t=1 (1+i)t
Where CAP EX is the initial capital expenditures, O&M is the operations and management parameter, D is the decommissioning parameter, Et is the annual energy produced in kWh, i is the discount or WACC rate, and t is the annual time step, where t = {1 ...T }.
5.3.6 Validation
Capital Cost Model
Since capital expenditures represent the largest proportion of costs, and much of the ancillary costs are fixed fractions of capital costs, it is critically important to validate our model against stated offshore wind farm costs worldwide. To do so, we collected data from http://www.4coffshore.com/ and http://www.lorc.dk/offshore-wind-farms-map/statistics on stated capital costs and designs for wind farms that are in construction or currently operational. We constrained the data collection to only those employing 3.6 MW and 5.0 MW turbines, for which we have provided default data with the InVEST model. Stated capital costs gathered from 4Coffshore were inflated to 2012 $US using their supplied financial close information as the basis for when the cost estimate was collected. To generate predictions, the design of each farm was input into the InVEST model using appropriate default cost parameters for all components. Most farms have their own electrical transmission equipment, though some deepwater farms are beginning to used centralized offshore substations that aggregate energy for transport from multiple farms. To predict electrical transmission costs for these farms, it was first necessary to estimate the cost of the entire offshore substation and then attribute a prorated
5.3. The model 51 InVEST User Guide, Release 2.5.6 capital cost to each farm based on their relative contribution to exported energy capacity. For example, an offshore substation with a 800 MW export capacity that is connected to Farm A (200 MW) and Farm B (600 MW) would contribute 25% of capital costs to Farm A and 75% to Farm B. The results of our validation show a very strong correlation between predictions and stated capital costs for 3.6 MW and 5.0 MW turbines using the default data (see Figure 5.6).
Figure 5.5: Predicted capital costs versus stated capital costs.
5.3.7 Limitations and Simplifications
Energy Production
The quality of wind input data determines the accuracy of model results. So, users need to understand the quality of wind input data for proper interpretation of the model results. The default wind input data are more appropriate for global and regional scale applications at 4 or 60 minutes spatial resolution. Harvested wind energy indicates the averaged energy output for a given period based on the output power curve of a wind turbine. Users may want to consider additional technology-specific information, such as device availability, power conversion efficiency, and directional factors by applying adjustment factors to the harvested energy output.
Energy Valuation
As the validation section demonstrates, the model and the default data reliably predict capital costs using the supplied inputs. Revenues are linked to energy production and a user-entered price. Currently the model is not capable of han- dling a price schedule with time-variant energy pricing. Inflation can be incorporated into the discount rate parameter, but more elaborate price changes are currently outside of the model’s capability. More reliable cost projections over space could likely be attained by:
5.3. The model 52 InVEST User Guide, Release 2.5.6
• Creating a foundation cost function that accounts for higher costs in deeper waters • Having installation costs vary as a function of bottom geology These are features that are being explored for subsequent model updates. The model is amenable to producing valuation outputs for floating turbines, but was not designed specifically for this task. To produce outputs, the user needs to input reasonable values for depth and distance constraints as well as “foundation” costs equal to the unit cost of the aggregate equipment needed to float a turbine. The electrical transmission model was derived using technologies that are suitable to roughly 60 meters depth and 200 kilometers distance from shore and will likely produce less accurate cost projections outside of those bounds. Net present value results are pre-tax net earnings.
5.4 Data Needs
5.4.1 Required Inputs
1. Workspace (required). Select a folder to be used as your workspace. If the folder you select does not exist, a new one will be created. This folder will contain the rasters produced by this model. If datasets already exist in this folder, they will be overwritten. The output will be contained in an folder named output inside the workspace directory. 2. Wind Data Points (required). A binary file that represents the wind input data (Weibull parameters). This parameter box should point to one of two files provided by our model. These files are found in the WindEnergy- input direction inside the InVEST installation directory. • Global Data: GLobal_EEZ_WEBPAR_90pct_100ms.bin • East Coast of the US: ECNA_EEZ_WEBPAR_Aug27_2012.bin for finer resolution of that area. 3. Area Of Interest (AOI) An optional polygon shapefile that defines the area of interest. The AOI must be projected with linear units equal to meters. If the AOI is provided it will clip and project the outputs to that of the AOI. The Distance inputs are dependent on the AOI and will only be accessible if the AOI is selected. If the AOI is selected and the Distance parameters are selected, then the AOI should also cover a portion of the land polygon to calculate distances correctly. An AOI is required for valuation. 4. Bathymetry (DEM) A raster dataset for the elevation values in meters of the area of interest. The DEM should cover at least the entire span of the area of interest and if no AOI is provided then the default global DEM should be used. 5. Land Polygon for Distance Calculation A polygon shapefile that represents the land and coastline that is of interest. For this input to be selectable the AOI must be selected. The AOI should also cover a portion of this land polygon to properly calculate distances. This coastal polygon, and the area covered by the AOI, form the basis for distance calculations for wind farm electrical transmission. This input is required for masking by distance values and for valuation. 6. Global Wind Energy Parameters A CSV file that holds wind energy model parameters for both the biophysical and valuation modules. These parameters are defaulted to values that are reviewed in the The Model section of this guide. We recommend careful consideration before changing these values. 7. Results Suffix A String that will be added to the end of the output file paths. 8. Turbine Type A CSV file that contains parameters corresponding to a specific turbine type. The InVEST package comes with two turbine model options, 3.6 MW and 5.0 MW. You may create a new turbine class (or modifying existing classes) by using the existing file format conventions and filling in your own parame- ters. It is recommended that you do not overwrite the existing default CSV files. These files are found in the WindEnergyinput direction inside the InVEST installation directory and named
5.4. Data Needs 53 InVEST User Guide, Release 2.5.6
• 3.6 MW: 3_6_turbine.csv • 5.0 MW: 5_0_turbine.csv 9. Minimum Depth for Offshore Wind Farm Installation (m) A floating point value in meters for the minimum depth of the offshore wind farm installation. 10. Maximum Depth for Offshore Wind Farm Installation (m) A floating point value in meters for the maximum depth of the offshore wind farm installation. 11. Minimum Distance for Offshore Wind Farm Installation (m) A floating point value in meters that represents the minimum distance from shore for offshore wind farm installation. Required for valuation. 12. Maximum Distance for Offshore Wind Farm Installation (m) A floating point value in meters that represents the maximum distance from shore for offshore wind farm installation. Required for valuation.
Valuation
13. Cost of the Foundation Type (millions of dollars) A floating point number for the unit cost of the foundation type (in millions of dollars). The cost of a foundation will depend on the type of foundation selected, which itself depends on a variety of factors including depth and turbine choice. 14. Number Of Turbines An integer value indicating the number of wind turbines per wind farm. 15. Price of Energy per Kilowatt Hour ($/kWh) The price of energy per kilowatt hour. 16. Discount Rate The discount rate reflects preferences for immediate benefits over future benefits. Enter in decimal form (Ex: 1% as 0.01, 100% as 1.0). 17. Grid Connection Points An optional CSV file with grid and land points to determine energy transmission cable distances from. Each point location is represented as a single row with columns being ID, TYPE, LATI, and LONG. The LATI and LONG columns indicate the coordinates for the point. The TYPE column relates to whether it is a land or grid point. The ID column is a simple unique integer. The shortest distance between respective points is used for calculations. An example: ID TYPE LATI LONG 1 GRID 42.957 -70.786 2 LAND 42.632 -71.143 3 LAND 41.839 -70.394 18. Average Shore to Grid Distance (km) A number in kilometers that is only used if grid points are NOT used in valuation. When running valuation using the land polygon to compute distances, the model uses an average distance to the onshore grid from coastal cable landing points instead of specific grid connection points.
5.5 Interpreting Results
All output resolutions are based on the resolution of the supplied digital elevation model raster. When the resolution of the DEM exceeds the resolution of the wind data layers, pixel values are determined by using blinear interpolation. • carbon_emissions_tons.tif : a GeoTIFF raster file that represents tons of offset carbon emissions for a farm built centered on a pixel per year. • density_W_per_m2.tif : a GeoTIFF raster file that represents power density (W/m^2) centered on a pixel. • example_size_and_orientation_of_a_possible_wind_farm.shp: an ESRI shapefile that represents the outer boundary of a sample windfarm. The position of this polygon is random and is meant to give the user a sense of scale of the potential wind farm.
5.5. Interpreting Results 54 InVEST User Guide, Release 2.5.6
• harvested_energy_MWhr_per_yr.tif : a GeoTIFF raster file that represents the annual harvested energy from a farm centered on that pixel. • levelized_cost_price_per_kWh.tif : a GeoTIFF raster file that represents the unit price of energy that would be required to set the present value of the farm centered at that pixel equal to zero. • npv_US_millions.tif : a GeoTIFF raster file that represents the net present value of a farm centered on that pixel. • wind_energy_points.shp: an ESRI Shapefile that summarizes the above outputs for each point...
5.6 Data Sources
5.6.1 Energy Output Data
• Wind time series data: NOAA’s National Weather Service provides hindcast reanalysis results for wind time series; http://polar.ncep.noaa.gov/waves/index2.shtml. The spatial resolution of the model results ranges from 4 to 60 minutes depending on the global and regional grid systems. The model outputs have been saved at 3-hour interval from 1999 to the present. The model results have been validated with ocean buoy data at many locations and provide good quality wind information. • Water depth: NOAA’s National Geophysical Data Center (NGDC) provides global bathymetry data with various spatial resolutions at http://www.ngdc.noaa.gov/mgg/bathymetry/relief.html. – ETOPO1 is a 1 arc-minute global relief model of Earth’s surface that integrates land topography and ocean bathymetry. It was built from numerous global and regional data sets, and is available in “Ice Surface” (top of Antarctic and Greenland ice sheets) and “Bedrock” (base of the ice sheets) versions. NGDC also provides regional and other global bathymetry datasets. • LORC knowledge provides the parameter information of offshore wind turbines that are currently operating in the world. http://www.lorc.dk/offshore-wind-farms- map/list?sortby=InstalledCapacity&sortby2=&sortorder=desc
5.6.2 Valuation
Data sources are largely cited above, except for figures that were derived from press releases. Press releases were found by an exhaustive Google keyword search on “offshore wind energy” contract and several variants of that theme. All costs were recorded and inflated in their original currency and exchanged to $US at the spot rate on March 30th, 2012.
5.7 Running the Model
To run the wind energy model, navigate to the “Wind Energy” application under the windows Start Menu found in All Programs->InVEST{version}->Marine. The user interface will indicate the required and optional input arguments as described in the Data Needs section above. Click the Run button to start the model. A successful run will be indicated in the window and a file explorer will open containing the results. If you encounter any errors please email the output log to [email protected].
5.6. Data Sources 55 InVEST User Guide, Release 2.5.6
5.8 References
AWS Truewind. 2010. New York’s Offshore Wind Energy Development Potential in the Great Lakes. Feasibility Study for New York State Energy Research and Development Authority. Blanco, M. 2009. The Economics of Wind Energy. Renewable and Sustainable Energy Reviews, 13, 1372-82. Boccard, N. 2010. Economic Properties of Wind Power: A European Assessment. Energy Policy, 38, 3232-3244. Carbon Trust. 2008. Offshore Wind Power: Big Challenge, Big Opportunity. Report on behalf of the Government of the United Kingdom. Celik, A. N. 2003. A statistical analysis of wind power density based on the Weibull and Rayleigh models at the southern of Turkey. Renewable Energy 29:509-604. Elliott, D. L., C. G. Holladay, W. R. Barchet, H. P. Foote, and W. F. Sandusky. 1986. Wind energy resource atlas of the United States. DOE/CH 10093-4. Solar Technical Information Program, Richland, Washington. Global Wind Energy Council (GWEC). 2013. Global Wind Statistics, 2012. Accessed at: http://www.gwec.net/wp- content/uploads/2013/02/GWEC-PRstats-2012_english.pdf Hsu, S. A., E. A. Meindl, and D. B. Gilhousen. 1994. Determining the power-law wind-profile exponent under near-neutral stability conditions at sea. Journal of applied meteorology 33:757-765. Jafarian, M., and A. M. Ranjbar. 2010. Fuzzy modeling techniques and artificial neural networks to estimate annual energy output of a wind turbine. Renewable Energy 35:2008-2014. Kaiser, M. and B. Snyder. 2012. Offshore wind capital cost estimation in the U.S. Outer Continental Shelf: A reference class approach. Marine Policy, 36, 1112-1122 Levitt, A., Kempton, W., Smith, A., Musial, W., and J. Firestone. 2011. Pricing offshore wind energy. Energy Policy, 39, 6408-6421. Lorc Knowledge. 2012. List of offshore wind farms. http://www.lorc.dk/offshore-wind-farms-map/list Accessed at December 31, 2012. Manwell, J. F., J. G. Mcgowan, and A. L. Rogers. 2009. Wind energy explained: Theory, design and application. John Wiley & Sons Ltd., West Sussex, United Kingdom. National Oceanic and Atmospheric Administration. 1976. U. S. Standard Atmosphere. NOAA- S/T76-1562, Wash- ington, DC. Pallabazzer, R. 2003. Provisional estimation of the energy output of wind generators. Renewable Energy 29:413-420. UMaine. 2011. Maine deepwater offshore wind report. http://www.deepcwind.org/docs/OfficialOffshoreWindReport- 22311.pdf
5.8. References 56 CHAPTER SIX
COASTAL VULNERABILITY MODEL
6.1 Summary
Faced with a changing climate and a growing intensity of human activities, coastal communities must better understand how modifications of the biological and physical environment (i.e. direct and indirect removal of natural habitats for coastal development) can affect their exposure to storm-induced erosion and flooding (inundation). The InVEST Coastal Vulnerability model produces a qualitative estimate of such exposure in terms of a Vulnerability Index, which differentiates areas with relatively high or low exposure to erosion and inundation during storms. By coupling these results with global population information, the model can show areas along a given coastline where humans are most vulnerable to storm waves and surge. The model does not take into account coastal processes that are unique to a region, nor does it predict long- or short-term changes in shoreline position or configuration. Model inputs, which serve as proxies for various complex shoreline processes that influence exposure to erosion and inundation, include: a raster representing population distribution, a polyline with attributes about local coastal geo- morphology along the shoreline, polygons representing the location of natural habitats (e.g., seagrass, kelp, wetlands, etc.), rates of net sea-level change, a depth contour that can be used as an indicator for surge level (the default contour is the edge of the continental shelf), a digital elevation model (DEM) representing the topography of the coastal area, and values of highest observed wind speed and wave power. Outputs can be used to better understand the relative contributions of these different model variables to coastal exposure and to highlight the protective services offered by natural habitats to coastal populations. This information can help coastal managers, planners, landowners and other stakeholders identify regions of greater risk to coastal hazards, which can in turn can better inform development strategies and permitting. As the results provide a qualitative representation of erosion and inundation risks rather than quantifying shoreline retreat or inundation limits, this is a “Tier 0” model within the InVEST framework.
6.2 Introduction
Coastal regions are constantly subject to the action of ocean waves and storms and naturally experience erosion and inundation over various temporal and spatial scales. Further, coastal erosion and inundation pose a threat to human populations, activities and infrastructure, especially within the context of a changing climate and increasing coastal populations. Moreover, these increases in anthropogenic pressure can lead to the loss and degradation of coastal ecosystems and their ability to provide protection for humans during storms. Thus, it is important to understand the role of various biological and geophysical factors in increasing or decreasing the threat of coastal erosion and inundation in order to better plan for future development. In particular, it is important to know how natural habitats can mitigate the forces responsible for coastal erosion and inundation so that management actions might best preserve the protective services provided by coastal ecosystems. A number of models estimate the vulnerability of coastal regions to long-term sea level rise, erosion and inundation based on geophysical characteristics (Gornitz et al. 1991, Hammar-Klose and Thieler 2001, Cooper and McLaughlin 1998). There are also methods to qualitatively estimate the relative role natural habitats play in reducing the risk
57 InVEST User Guide, Release 2.5.6 of erosion and inundation of particular areas (WRI 2009, Bush et al. 2001). However, few models map the relative vulnerability of coastal areas to erosion and inundation based on both the geophysical and natural habitat characteristics of a region. It is our aim to fill that gap with the Coastal Vulnerability model. The Coastal Vulnerability model produces a map of the location and size of human settlements as well as a qualitative index of coastal exposure to erosion and inundation. The model does not value directly any environmental service, but ranks sites as having a relatively low, moderate or high risk of erosion and inundation. It is relatively simple to use and quick to run, and it can be applied in most regions of the world with data that are, for the most part, relatively easy to obtain. Model outputs are easy to understand and the spatial coverage of the results allows these outputs to be overlaid with other spatial information for users to perform further analysis as they see fit. By showing the areas where coastal populations are threatened and highlighting the relative role of natural habitat at reducing exposure, the model can be used, in a simple way, to investigate how some management action or land use change can affect the exposure of human populations to erosion and inundation.
6.3 The model
The InVEST Coastal Vulnerability model produces a coastal population raster and a Vulnerability Index raster. The coastal population raster shows the distribution of human population density within the coastal region of interest. The Vulnerability Index raster contains ranks of the relative exposure of coastlines and communities to erosion and inunda- tion caused by large storms within the same coastal region of interest. These maps are constructed using a population raster and seven bio-geophysical variables that represent the natural biological and geomorphic characteristics of a region, the amount of expected net sea-level rise and the relative wind and wave forcing associated with storms. As the model assesses the relative vulnerability within the domain of interest, model outputs are relevant when computed for a relatively large and/or non-uniform coastal region. Coupled together, the population and Vulnerability Index rasters can be used to create maps that show the relative vulnerability of human populations to coastal storms.
6.3.1 How it works
The model creates the population and exposure index maps using a spatial representation (raster) of population and spatial representations (shapefiles and rasters) of seven bio-geophysical variables: 1. Geomorphology 2. Relief 3. Natural habitats (biotic and abiotic) 4. Net sea level change 5. Wind Exposure 6. Wave Exposure 7. Surge potential depth contour The outputs of the model are rasters along the shoreline in the coastal region of interest, which have a spatial reso- lution defined by the user (≥ 250 meters). These rasters contain a number of indices and rankings of input variables (described below) and can be used to the create maps that fit the users needs. Below are details describing the model variables and how the output rasters are created.
Social Exposure
When estimating the exposure of coastlines to erosion and inundation due to storms, it is important to consider the population of humans that will be subject to those coastal hazards. The Coastal Vulnerability model extracts population
6.3. The model 58 InVEST User Guide, Release 2.5.6 values along the shoreline at discrete segments with the user-specified spatial resolution. To obtain this raster showing the estimated number of people residing on a coastal area, the model overlays a raster containing population values pulled from a user-defined radius (see coastal neighborhood) at each grid cell with the rasterized shoreline. The model then assigns each discrete shoreline segment a population value by extracting the population value from the grid cell that overlaps the shoreline segment. As a part of the InVEST download package, a global population raster is provided with population values obtained from country level census data. As is the case with all input data, the user may provide their own population raster if they have more accurate, local information.
Vulnerability Index
The model computes the physical exposure index by combining the ranks of the seven biological and physical variables at each shoreline segment. Ranks vary from very low exposure (rank=1) to very high exposure (rank=5), based on a mixture of user- and model-defined criteria (see Table 4.1). This ranking system is based on methods proposed by Gornitz et al. (1990) and Hammar-Klose and Thieler (2001).
Table 4.1
Rank Very Low Low Moderate High Very High Vari- 1 2 3 4 5 able Geo- Rocky; high Medium cliff; Low cliff; glacial drift; Cobble Barrier beach; mor- cliffs; fjord; indented coast, alluvial plain, beach; sand beach; phol- fiard, seawalls bulkheads and small revetments, rip-rap estuary; mud flat; delta ogy seawalls walls lagoon; bluff Relief <=20th <=40th Percentile <=60th Percentile <=80th >80th Percentile Percentile Percentile Natu- Coral reef; High dune; marsh Low dune Seagrass; No habitat ral mangrove; kelp Habi- coastal forest tats Sea Net decrease -1 to +1 Net rise Level Change Wind <=20th <=40th Percentile <=60th Percentile <=80th >80th Expo- Percentile Percentile Percentile sure Wave <=20th <=40th Percentile <=60th Percentile <=80th >80th Expo- Percentile Percentile Percentile sure Surge <=20th <=40th Percentile <=60th Percentile <=80th >80th Poten- Percentile Percentile Percentile tial Table 4.1: List of Bio-Geophysical Variables and Ranking System for Coastal Exposure. The model calculates the exposure index EI for each shoreline segment as the geometric mean of all the variable ranks:
1/7 EI = (RGeomorphologyRRelief RHabitatsRSLRRW indExposureRW aveExposureRSurge) (6.1)
6.3. The model 59 InVEST User Guide, Release 2.5.6
or more generally:
n !1/n Y EI = Ri (6.2) i=1
th where Ri represents the ranking of the i bio-geophysical variable that is provided by the user to calculate EI. In addition to mapping the exposure index, the model computes an erosion index ErI as:
1/3 ErI = (RGeomorphologyRHabitatsRW aveExposure) (6.3)
We designed this additional output to invite users to explore the different ways in which model results vary with different combinations of variables. Here, we have assumed that the most important factors affecting erosion are geomorphology, natural habitats and wave height/period. The erosion index provides an example for how users can similarly create their own index by defining a unique combination of these rankings. In the remainder of this section, we will provide a more detailed descriptions of the variables presented in Table 4.1.
Geomorphology
Rocky cliffs are less prone to erosion and inundation than bluffs, beaches or deltas. Consequently, a relative ranking of exposure scheme based on geomorphology similar to the one proposed by Hammar-Klose and Thieler (2001) has been adopted. Supplied in Appendix A is a definition of the terms used in this classification, which applies mostly to the North American continent. This classification will be expanded to cover more regions of the world in later versions of this model. In addition, we included structures in this list of features because they are present along most developed coasts. If the user’s geomorphology source has more categories than the ones presented in Table 4.1, it is left to the users discretion to reclassify their data to match the provided ranking system, as explained in the Data needs section, and in Appendix B. It is recommend however, that the user include shore parallel hard structures (seawalls, bulkheads, etc) in this classification and that they apply a low to moderate rank (1-3), depending on their characteristics. For example, a large, concrete seawall should be assigned a rank 1 as they are typically designed to prevent inundation during storm events and are designed to withstand damage or failure during the most powerful storms. It is recommended that low revetments or riprap walls be assigned a rank of 3 as they do not prevent inundation and may fail during extreme events. The model requires a polyline shapefile that runs along the coastline of interest. This shapefile must be discretized into different segments where the geomorphology rank changes. The attribute table of this shapefile requires a field called RANK of type ‘Short Integer’, which contains the numeric rank (1-5) of the geomorphology type along each particular segment.
Relief
Sites that are, on average, at greater elevations above Mean Seal Level (MSL) are at a lower risk of being inundated than areas at lower elevations. Relief is defined in the model as the average elevation of the coastal land area that is within a user-defined radius (default = 5 km) from each shore segment of the discretized shoreline. This resolution was chosen because of the relative coarseness of most freely available terrestrial digital elevation models (DEMs). For this variable, the model requires an ESRI GRID DEM that covers the area of interest. As a part of the InVEST download package, a global elevation DEM is provided. The accuracy and quality of this dataset varies. If the user has access to a more local, accurate data source, it is encourages that this source is used rather than the provided DEM.
6.3. The model 60 InVEST User Guide, Release 2.5.6
Natural Habitats
Natural habitats (marshes, seagrass beds, mangroves, coastal dunes) play a vital role in decreasing coastal hazards that harm shorelines and coastal community. For example, large waves break on coral reefs before reaching the shoreline, mangroves and coastal forests dramatically reduce wave height in shallow waters, and decrease the strength of wave- and wind-generated currents, seagrass beds and marshes stabilize sediments and encourage the accretion of nearshore beds as well as dissipate wave energy. On the other hand, beaches with little to no biological habitats or sand dunes offer little protection to erosion and inundation. The ranking proposed in Table 4.1 is based on the fact that fixed and stiff habitats that penetrate the water column (e.g., coral reefs, mangroves) and sand dunes are the most effective in protecting coastal communities. Flexible and seasonal habitats, such as seagrass, reduce flows when they can withstand their force, and encourage accretion of sediments. Therefore, these habitats receive a lower ranking than fixed habitats. Once again, it is left to the user’s discretion to separate sand dunes into high and low categories. It is suggested, however, that since category 4 hurricanes can create a 5m surge height, 5m is an appropriate cut-off value to separate high (>5m) and low (<5m) dunes. If the user has local knowledge about which habitats and dune elevations provide the better protection in their area of interest, they are free to deviate from these recommendations for their application. To compute a Natural Habitat exposure rank for a given shoreline segment, the model determines whether a certain class of natural habitat (Table 4.1) is within a user-defined search radius from the segment. (See Section 2 and Appendix B for a description of how the model processes natural habitat input layers.) When all N habitats fronting that segment have been identified, the model creates a vector R that contains all the ranks Rk, 1 ≤ k ≤ N, associated with these habitats, as defined in Table 4.1. Using those rank values, the model computes a final Natural Habitat exposure rank for that segment with the following formulation: v u N u N 2 X 2 N 2 RHab = 4.8 − 0.5t(1.5 max(5 − Rk)) + ( (5 − Rk) − max(5 − Rk)) ) (6.4) k=1 k=1 k=1
where the habitat that has the lowest rank is weighed 1.5 times higher than all other habitats that are present near a segment. The final ranking values vary between a maximum of 4 when a segment is solely fronted by kelp or seagrass, to a minimum of 1.025 when it is fronted by a mangrove and coastal forests, a seagrass bed and a coral reef. This formulation allows us to maximize the accounting of the protection services provided by all natural habitats that front a shoreline segment. In other words, it ensures that segments that are fronted or have only one type of habitat (e.g., high sand dune) are more exposed than segments with more than one habitat (e.g., coral reefs and high sand dune). See Appendix B for a detailed account of all possible final rank values that can be obtained with equation (6.4). To include this variable in the exposure index calculation, the model requires separate polygon shapefiles representing each natural habitat type within the area of interest, along with a csv file that contains the name, rank and search radius of each of the shapefile. We present a complete description of the requirements for this variable and instructions on how to prepare this variable for the model in Appendix B.
Net Sea-Level Change
The relative net sea level rise/decrease along the coastline of a given region is the sum of global SLR, local SLR (eustatic rise) and local land motion (isostatic rise). As indicated by Gornitz (1990), relative rise values between -1 and +1 do not change current erosion or inundation trends, as they can be considered to be within modeling and measurement error range. In contrast, values smaller than -1 decrease the exposure, while values above +1 increase the exposure. The model takes either a polygon shapefile where polygons delineate the extents of a uniform sea level change, or a point shapefile where the points carry the recorded sea level change. In either case, the model will look for a field named ‘Trend’, which is the yearly rate of sea level change. Please consult Appendix B for suggestions of how to create this input.
6.3. The model 61 InVEST User Guide, Release 2.5.6
Wind Exposure
Strong winds can generate high surges and/or powerful waves if they blow over an area for a sufficiently long period of time. The wind exposure variable ranks shoreline segments based on their relative exposure to strong winds. We compute this ranking by computing and mapping the Relative Exposure Index (REI; Keddy, 1982). This index is computed by taking the time series of the highest 10% wind speeds from a long record of measured wind speeds, dividing the compass rose (or the 360 degrees compass) into 16 equiangular sectors and combining the wind and fetch (distance over which wind blows over water) characteristics in these sectors as:
16 X REI = UnPnFn (6.5) n=1 where: th • Un is the average wind speed, in meters per second, of the 10% wind speeds in the n equiangular sector th • Pn is the percent of all wind speeds in the record of interest that blow in the direction of the n sector th • Fn is the fetch distance, in meters, in the n sector For a given coastline segment, the model estimates fetch distances over each of the 16 equiangular sectors, with an accuracy of 1km, by using the model developed by Finlayson (2005). Please note that, in this model, wind direction is the direction winds are blowing FROM, and not TOWARDS. If the user provides their own data, the user must ensure that the data matches this convention before applying those data to this model.
Wave Exposure
The relative exposure of a reach of coastline to storm waves is a qualitative indicator of the potential for shoreline erosion. A given stretch of shoreline is generally exposed to either oceanic or locally-generated wind-waves. Also, for a given wave height, waves that have a longer period have more power than shorter waves. Coasts that are exposed to the open ocean generally experience a higher exposure to waves than sheltered regions because winds blowing over a very large distance, or fetch, generate larger waves. Additionally, exposed regions experience the effects of long period waves, or swells, that were generated by distant storms.
The model estimates the relative exposure of a shoreline segment to waves Ew by assigning it the maximum of the o l weighted average power of oceanic waves, Ew‘andlocallywind − generatedwaves, : math :‘Ew:
o l Ew = max(Ew,Ew) (6.6)
For oceanic waves, the weighted average power is computed as:
16 o X o o Ew = H[Fk]Pk Ok (6.7) k=1
where H[Fk] is a heaviside step function for all of the 16 wind equiangular sectors k. It is zero if the fetch in that direction is less than 50km, and 1 if the fetch is greater than or equal to 50km: ( 0 if Fk < 50km H[Fk] = (6.8) 1 if Fk ≥ 50km
6.3. The model 62 InVEST User Guide, Release 2.5.6
In other words, this function only considers angular sectors where oceanic waves (assuming sheltered water bodies have fetch lengths less than 50km) have the potential to reach the shoreline in the evaluation of oceanic wave exposure. o o o Further, Pk Ok is the average of the highest 10% wave power values (Pk ) that were observed in the direction of the o angular sector k, weighted by the percentage of time (Ok) when those waves were observed in that sector. For all waves in each angular sector, wave power is computed as: 1 P = H2T (6.9) 2
where P [kW/m] is the wave power of an observed wave with a height H[m] and a period T [s]. l For locally wind-generated waves, Ew is computed as: 16 l X l l Ew = PkOk (6.10) k=1
which is the sum over the 16 wind sectors of the wave power generated by the average of the highest 10% wind speed l l values Pk that propagate in the direction k, weighted by the percent occurrence Ok of these strong wind in that sector. The power of locally wind-generated waves is estimated with Equation (6.9). The wave height and period of the locally generated wind-waves are computed for each of the 16 equiangular sectors as:
h −4 0.79 i0.572 1.14 2.14·10 Fe H = He∞ tanh 0.343de tanh tanh(0.343de1.14) (6.11) h −7 1.45 i0.187 2.01 2.77·10 Fe T = Te∞ tanh 0.1de tanh tanh(0.1de2.01)
where the non-dimensional wave height and period He∞ and Te∞ are a function of the average of the highest 10% wind 2 2 speed values U[m/s] that were observed in in a particular sector: He∞ = 0.24U /g, and Te∞ = 7.69U /g, and where the non-dimensional fetch and depth, Fe∞ and de∞, are a function of the fetch distance in that sector F [m] and the 2 2 2 average water depth in the region of interest d[m]: Fe∞ = gF/U , and Te∞ = gd/U . g[m/s ] is the acceleration of gravity. This expression of wave height and period assumes fetch-limited conditions, as the duration over which the wind speed,:math:U, blows steadily in the direction of the fetch, F (USACE, 2002; Part II Chap 2). Hence, model results might over-estimate wind-generated waves characteristics at a site. Since sheltered areas of the coast (areas that are within embayments or sheltered from oceanic waves by geomorphic o l features) are not exposed to oceanic waves (Ew = 0) the relative exposure to waves is simply Ew = Ew. In order to differentiate between exposed and sheltered areas , the model uses a fetch filter; segments for which two or more of the 16 fetches do not exceed a user-defined threshold distance are assumed to be sheltered. As a part of the InVEST download package, a shapefile with default wind and wave data compiled from 6 years of WAVEWATCH III (WW3, Tolman (2009)) model hindcast reanalysis results is provided. As discussed in the previous section, for each of the 16 equiangular wind sector, the average of the highest 10% wind speed, wave height and wave power have been computed. If the user wishes to use another data source, they must use the same statistics of wind and wave (average of the highest 10% for wind speed, wave height and wave power) in order to produce meaningful results. Further, these data must be contained in a point shapefile with the same attribute table as the WW3 data provided.
Surge Potential
Storm surge elevation is a function of wind speed and direction, but also of the amount of time wind blows over relatively shallow areas. In general, the longer the distance between the coastline and the edge of the continental
6.3. The model 63 InVEST User Guide, Release 2.5.6 shelf at a given area during a given storm, the higher the storm surge. Unless a user decides to specify a certain depth contour appropriate to their region of interest, the model estimates the relative exposure to storm surges by computing the length of the continental shelf fronting an area of interest (otherwise, it computes the distance between the shoreline and the user-specified contour). For hurricanes, a better approximation might be made by considering the distance between the coastline and the 30 meters depth contour (Irish and Resio 2010). The model assigns a distance to all segments within the area of interest, even to segments that seem sheltered because they are too far inland, protected by a significant land mass, or on a side of an island that is not exposed to the open ocean. Consequently, the user can provide a maximum distance threshold over which shoreline segments within the area of interest will be deemed at low-risk of exposure to storm surge (see Data needs section). We provide an example of how to estimate this distance in Appendix B.
6.4 Limitations and Simplifications
Beyond technical limitations, the Exposure Index also has theoretical limitations. One of the main limitations is that the dynamic interactions of complex coastal processes occurring in a region are overly simplified into the geometric mean of seven variables and exposure categories. We do not model storm surge or wave field in nearshore regions. More importantly, the model does not take into account the amount and quality of habitats, and it does not quantify the role of habitats are reducing coastal hazards. Also, the model does not consider any hydrodynamic or sediment transport processes: it has been assumed that regions that belong to the same broad geomorphic exposure class behave in a similar way. Additionally, the scoring of exposure is the same everywhere in the region of interest; the model does not take into account any interactions between the different variables in Table 4.1. For example, the relative exposure to waves and wind will have the same weight whether the site under consideration is a sand beach or a rocky cliff. Also, when the final Exposure Index is computed, the effect of biogenic habitats fronting regions that have a low geomorphic ranking are still taken into account. In other words, we assume that natural habitats provide protection to regions that are protected against erosion independent of their geomorphology classification (i.e. rocky cliffs). This limitation artificially deflates the relative vulnerability of these regions, and inflates the relative vulnerability of regions that have a high geomorphic index. The other type of model limitations is associated with the computation of the wind and wave exposure. Because our intent is to provide default data for users in most regions of the world, we had to simplify the type of input required to compute wind and wave exposure. For example, we computed storm wind speeds in the WW3 wind database that we provide by taking the average of winds speeds above the 90th percentile value, instead of using the full time series of wind speeds. Thus we do not represent fully the impacts of extreme events. Also, we estimate the exposure to oceanic waves by assigning to a coastal segment the waves statistics of the closest WW3 grid point. This approach neglects any 2D processes that might take place in nearshore regions and that might change the exposure of a region. Similarly, we compute exposure in sheltered region by combining the average depth near a particular segment to the wind speed and direction in a sector, instead of modeling the growth and evolution of wind waves near that segment. Consequently, model outputs cannot be used to quantify the exposure to erosion and inundation of a specific coastal location; the model produces qualitative outputs and is designed to be used at a relatively large scale. More importantly, the model does not predict the response of a region to specific storms or wave field and does not take into account any large-scale sediment transport pathways that may exist in a region of interest.
6.5 Data needs
The model uses an interface to input all required and optional data, as outlined in this section. It outputs a HTML file with a map of the area over which the model has been run, and three histograms showing the vulnerability of the population living near the coast, the exposure of coastal segments near urban centers, and the exposure of the whole coast. To compute the Exposure Index the user has the option of uploading any or all of the variables in Table 4.1, with the exception of the wind-wave input layer and the bathymetry: the model will not run unless a wind-wave input layer and DEM have been uploaded.
6.4. Limitations and Simplifications 64 InVEST User Guide, Release 2.5.6
Below, we outline the options that we offer to users in the interface, and the content and format of the required and optional input data that the model uses. we provide more information on how to fill the input interface, or on how to obtain data in Appendix B. 1. Output area. Specify whether all or only the sheltered shoreline segments appear in the output. This option has no effect on the computation performed by the model, and only affects the shore segments that appear in the output files. 2. Workspace Location (required). The user is required to specify a workspace directory path. It is recommended to create a new directory for each run of the model. The model will create an “intermediate” and an “output” directory within this workspace. The “intermediate” directory will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “output” directory. Name: Path to a workspace directory. Avoid spaces. Sample path: \InVEST\coastal_vulnerability
3. Area of Interest (AOI, required). Users must create a polygon feature layer that defines the Area of Interest (AOI). An AOI instructs the model where to clip the Land Polygon input data (inputs #2-3) in order to define the spatial extent of the analysis. The model uses the AOI’s projection to set the projection for the sequential intermediate and output data layers and must have a WGS84 datum. In order to allocate wind and wave infor- mation from the Wave Watch 3 data (WW3), this AOI must also overlap one or more of the provided WW3 points. If users are including the Surge Potential variable in the computation of the exposure index, the depth contour specified in the Coastal Vulnerability model must be specified, and the AOI must intersect that contour. If the AOI does not intersect that contour, the model will stop and provide feedback. Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\AOI_BarkClay.shp
4. Land Polygon (required). This input provides the model with a geographic shape of the coastal area of in- terest, and instructs it as to the boundaries of the land and seascape. A global land mass polygon shapefile is provided as default (Wessel and Smith, 1996), but other layers can be substituted. If users have a more accurate, local polygon shapefile representing land masses, they are encouraged to use this data rather than the provided shapefile. Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path (default): \InVEST\Base_Data\Marine\Land\global_polygon.shp
5. Bathymetry layer. (required) This input is used to compute the average depth along the fetch rays to determine the exposure of each shoreline segment (Table 4.1), and in the computation of surge potential. It should consist of depth information of bodies of water within the AOI as marked by the land polygon shapefile. Name: File can be named anything, but no spaces in the name File type: raster dataset Sample path: \InVEST\Base_Data\Marine\DEMs\claybark_dem
6. Layer value if path omitted (optional). Integer value between 1 and 5. If bathymetry is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
7. Relief (required). Digital Elevation Model (DEM). This input is used to compute the Relief ranking of each shoreline segment (Table 4.1). It should consist of elevation information covering the entire land polygon within the AOI. Focal statistics are computed on the input DEM within a range defined by the user (see Elevation averaging radius). The average of elevation values within this range is ranked relative to all other coastline
6.5. Data needs 65 InVEST User Guide, Release 2.5.6
segments within the AOI. Although the default raster for this layer is the same as for Bathymetry, each entry can refer to a separate raster, where one computes elevations above water, and the other below water. Name: File can be named anything, but no spaces in the name File type: raster dataset Sample path: \InVEST\Base_Data\Marine\DEMs\claybark_dem\hdr.adf
8. Layer value if path omitted (optional). Integer value between 1 and 5. If relief is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
9. Elevation averaging radius (meters, required). This input determines the radius around within which to compute the average elevation for relief. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 5000
10. Mean sea level datum (meters, required). This input is the elevation of Mean Sea Level (MSL) datum relative to the datum of the bathymetry layer that they provide. The model transforms all depths to MSL datum by subtracting the value provided by the user to the bathymetry. This input can be used to run the model for a future sea-level rise scenario.: Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 0
11. Smallest detectable feature (segment size in meters, required). This input determines the spatial resolution at which the model runs and the resolution of the output rasters. To run the model at 250 x 250 meters grid cell scale, users should enter “250”. A larger grid cell will yield a lower resolution, but a faster computation time (computation is in the order of :math:’O(n^3)’ with n being the number of rows or columns in the raster). Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 250
12. Rays per sector (required). Number of rays used to sample the ocean depth and land proximity within each of the 16 equiangular fetch sectors. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 1
13. Fetch Distance Threshold (meters, required). Used to determine if the current segment is enclosed by land. This input is used in conjunction with the average ocean depth and exposure proportion to differentiate sheltered and exposed shoreline segments.: Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 12000
14. Depth Threshold (meters, required). Used to determine if the current segment is surrounded by deep water. This input is used in conjunction with the fetch distance threshold and exposed segment to differentiate between sheltered and exposed shoreline segments.
6.5. Data needs 66 InVEST User Guide, Release 2.5.6
Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 0
15. Exposure proportion (meters, required). The model uses this input (between 0.0 and 1.0) to determine if shore segments are exposed or sheltered. This is done in four steps: (a) Compute the number of fetch rays (N) that correspond to the proportion N: :math:’segments over water * exposure proportion’ (b) Determine if the current segment is in deep water (at least N sectors project over water that is at least “depth threshold” meters) (c) Determine if the current segment is enclosed by land (at least N fetch rays have to be blocked by land, i.e. fetch distance is less than “ocean effect cutoff” meters). (d) Determine segment exposure: a shore segment is exposed if it is both in deep waters, and not enclosed by land (facing open water), otherwise, it is sheltered. In other words, if the fetch threshold is 12 km and the depth threshold is 5 m, and the exposure proportion is 0.8, the model will classify a segment as sheltered if less than 80% of the segments have a fetch distance lower than 12 km or the average depth along each fetch segment is less than 5 m. Name: A numeric text string (number between 0 and 1) File type: text string (direct input to the interface) Sample (default): 0.8 1. Oceanic effect cutoff (meters, required). Used as a threshold to determine if a shore segment is enclosed by land. See Exposure proportion, step 3. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 60000
2. Geomorphology: Shoreline Type (optional). This input is used to compute the Geomorphology ranking of each shoreline segment (Table 4.1). It does not have to match the land polyline input, but must resemble it as closely as possible. If it doesn’t, the model will try to match the coastlines using the coastal overlap parameter. Additionally, the polyline shapefile must have a field called “RANK” that identifies the various shoreline type ranks with a number from 1-5. More information on how to fill in this table is provided in Appendix B. Names: File can be named anything, but no spaces in the name File type: polyline shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\Geomorphology_BarkClay.shp
3. Layer value if path omitted (optional). Integer value between 1 and 5. If geomorphology is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
4. Coastal overlap (meters, required). Tolerance threshold in meters (that should be a multiple of cell size), to make 2 non-overlapping shorelines match. If the tolerance is twice the cell size, the model will be able to match shorelines that are 2 pixels off. If it is 4 times the cell size, the model will be able to match shorelines that are 4 pixels off, and so on. It’s used when the shoreline from geomorphology doesn’t overlap exactly with the shoreline from the land polygon shapefile. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 250
6.5. Data needs 67 InVEST User Guide, Release 2.5.6
5. Natural Habitat (optional). Directory that contains habitat layers. Users must store all Natural Habitats input layers that they have in a unique directory. The model uses these input layers to compute a Natural Habitat ranking for each shoreline segment. All data in this directory must be polyline or polygon shapefiles that depict the location of the habitats, and must be projected in meters. Additionally, each layer must end with an underscore followed by a unique alpha-numeric number. The model uses that number to match the habitat layer to the information that users provide in the CSV table (see next input). The model allows for a maximum of eight layers in this directory. Do not store any additional files that are not part of the analysis in this directory. The distance at which this layer will have a protective influence on coastline can be modified in the natural habitat CSV table (input 8). Name: Folder can be named anything, but no spaces in the name. Habitat layers in the folder must be of the form \93HabitatName_Number\94 (e.g., Coral_1), where the number is uniquely associated to that habitat layer. File type:a polygon shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\NaturalHabitat
6. Natural Habitat Layers CSV (Table optional).. Users must provide a summary table to instruct the model on the protective influence (rank) and distance of natural habitat. Use the sample table provided as a template since the model expects values to be in these specific cells. More information on how to fill this table is provided in Appendix B. Table Names: File can be named anything, but no spaces in the name File type: *.csv Sample: InVEST\CoastalProtection\Input\NaturalHabitat_WCVI.csv
22. Layer value if path omitted (optional). Integer value between 1 and 5. If natural habitats is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
23. Climatic forcing grid (optional). This input is used to compute the Wind and Wave Exposure ranking of each shoreline segment (Table 4.1). It consists of a point shapefile that contains the location of the grid points as well as wave and wind values that represent storm conditions at that location. If users would like to create such a file from their own own data, instructions are provided in Appendix B. Name: File can be named anything Format: point shapefile where each point has information about wind and wave measurements. Sample data set (default): \InVEST\CoastalProtection\Input\WaveWatchIII.shp
24. Layer value if path omitted (optional). Integer value between 1 and 5. If climatic forcing grid is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
6.5. Data needs 68 InVEST User Guide, Release 2.5.6
Sample (default): 500
25. Continental Shelf (optional). This input is a global polygon dataset that depicts the location of the continental margin. It must intersect with the AOI polygon (input #2). Names: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\continentalShelf.shp
26. Depth contour level (meters, optional). If no continental shelf is specified, the model will use the bathymetry data to trace a user-defined depth contour level ans use it instead of the edge of the continental shelf. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 150
27. Sea Level Rise (optional). Polygon Indicating Net Rise or Decrease. This input must be a polygon delineating regions within the AOI that experience various levels of net sea level change. It must have a field called “Trend”, which represents the rate of increase (mm/yr) of the sea level in a particular region according to Table 4.1. More information on how to create this polygon is provided in the Marine InVEST FAQ, and in Appendix B. Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) or point shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\SeaLevRise_WCVI.shp
28. Layer value if path omitted (optional). Integer value between 1 and 5. If sea level rise is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
29. Structures (optional). Polygon shapefile that contains the location of rigid structures along the coast. Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\Structures_BarkClay.shp
30. Layer value if path omitted (optional). Integer value between 1 and 5. If structures layer is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
31. Population Raster (optional). If provided, a raster grid of population is used to map the population size along the coastline of the AOI specified (input #4). A global population raster file is provided as default, but other population raster layers can be substituted. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: standard GIS raster file (ESRI GRID), with population values Sample data set (default): \InVEST\Base_Data\Marine\Population\global_pop\hdr.adf
32. Min. population in urban centers (required). Minimum population that has to live in the vincinity of a shore segment to be considered a urban center. The vincinity is defined in the next input, “coastal neighborhood”.
6.5. Data needs 69 InVEST User Guide, Release 2.5.6
Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 5000
33. Coastal neighborhood (radius in m, required). Radius defining the vincinity of a shore segment that is used to count the population living on or near the coast. Name: A numeric text string (positive integer) File type: text string (direct input to the interface) Sample (default): 150
34. Additional layer (optional). This additional layer can be any additional variable that users desire to add to the exposure index. It can be values of long-term shoreline change, for example. This layer must be a shapefile with [HOW DO YOU PROJECT TO SHORE, NIC?]. Once all segments have a value, we rank them according to quartile distribution. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: standard GIS raster file (ESRI GRID), with population values Sample data set (default): \InVEST\Base_Data\Marine\Population\global_pop
35. Layer value if path omitted (optional). Integer value between 1 and 5. If additional layer is omitted, replace all shore points for this layer with a constant rank value in the computation of the coastal vulnerability index. If both the file and value for the layer are omitted, the layer is skipped altogether. Name: A positive integer between 1 and 5. File type: text string (direct input to the interface) Sample (default): empty
6.6 Running the model
6.6.1 Setting up workspace and input directories
These directories will hold all input, intermediate and output data for the model.
Note: The word ‘path’ means to navigate or drill down into a directory structure using the Open Folder dialog window that is used to select GIS layers or Excel worksheets for model input data or parameters.
Exploring a project workspace and input data directory
The /InVEST/CoastalProtection directory holds the main working directory for the model and all other associated directories. Within the CoastalProtection directory there will be a sub-directory named ‘Input’. This directory holds most of the GIS and tabular data needed to setup and run the model. The following image shows the sample input (on the left) and base data (on the right) directory structures and accom- panying GIS data. It is recommend that this directory structure is used as a guide to organize workspaces and data. Refer to the screen-shots below for examples of directory structure and data organization.
6.6. Running the model 70 InVEST User Guide, Release 2.5.6
6.6.2 Creating a run of the model
The following example of setting up the Coastal Vulnerability (Tier 0) model uses the sample data provided with the InVEST download. The instructions and screen-shots refer to the sample data and directory structure supplied within the InVEST installation package. We expect that users will have location-specific data to use in place of the sample data. These instructions provide only a guideline on how to specify to the standalone 3.0 version of the model the various types of data needed and does not represent any site-specific model parameters. See the Data needs section for a more complete description of the data specified below. 1. On Windows 7, click on the “start” button to expand the start menu, then click on “All Programs” at the bottom. 2. Expand the folder which name starts with “InVEST”, and launch the model by clicking on “Coastal Vulnerabil- ity”. The model will show a user interface as shown in the next page. 3. Specify the area to appear in the output: sheltered shoreline segments, or both sheltered and exposed. 4. Specify the Workspace. Either enter the path to the workspace manually (the model will create it if it doesn’t already exist), or click on it from the navigation window (click on the Open Folder button on the right, default is InVEST/CoastalProtection). This is the directory in which the intermediate and final outputs will be stored when the model is run. 5. Specify the Area of Interest (AOI). The model requires an AOI, which is the geographic area over which the model will be run. This example refers to the sample shapefile AOI_BarkClay.shp supplied in In- VEST/CoastalProtection/Input. Users can create an AOI shapefile by following the Creating an AOI instructions in the FAQ section. 6. Specify the Land Polygon. The model requires a land polygon shapefile to define the shoreline for the analysis. A default path to the global sample data is supplied in the model window for users. 7. Specify the bathymetry (DEM raster) of the water in the AOI to be incorporated into Wave Exposure calcula- tions. It will be used to estimate wave height and associated period, for each of the 16 fetch angular sectors. 8. Specify the Relief Digital Elevation Model (DEM) raster. The model requires a DEM raster file to estimate average elevation landward of the coastal segment. The path of the default DEM file for the west coast of Vancouver Island is in InVEST/Base_Data/Marine/DEMs/claybark_dem.
6.6. Running the model 71 InVEST User Guide, Release 2.5.6
6.6. Running the model 72 InVEST User Guide, Release 2.5.6
8. Specify the elevation averaging radius (default is 5000, i.e. 5km). The model will average all the land elevations within this radius to compute relief. 9. Specify the mean sea level datum. The model can adjust the mean sea level relative to the datum of the bathymetry. Default is zero, and positive values indicate that the mean sea level datum is above the bathymetry’s datum. 10. Specify the size of the smallest detectable feature (Cell Size). The model requires a cell size for the raster analysis (default is 250 m). The model will not be able to distinguish details on rasters or shapefiles that are smaller than the size specified. 11. Specify the number of rays per sector: This is the number of rays the model uses to sample water depth and the presence of landmasses within each sector. Users may change this value by entering a new value directly into the text box. 12. Specify the maximum fetch distance. The model computes the fetch over a maximum distance to separate sheltered and exposed areas. The default value is 12,000 meters. The longer the distance, the longer the rays, and the slower the computation. 13. Specify a depth threshold (positive integer, or 0). The model uses the depth threshold to determine areas of shallow water. It is used to segregate exposed from sheltered shore segments. A value of zero (0) cancels the effect of this parameter. 14. Specify exposure proportion (real number between 0 and 1). The model requires a percentage of sectors to span either shallow or enclosed water bodies to classify the shore as sheltered. uses the depth threshold to determine areas of shallow water. It is used to segregate exposed from sheltered shore segments. A value of zero (0) cancels the effect of this parameter, as all water bodies will be considered deep. 15. The Oceanic effect cutoff is the maximum distance allowed to consider coastal segments enclosed. Set this value so that it is at least as long as the distance across areas that should be enclosed. 16. Specify the Geomorphology layer (optional). The model can use an optional polygon shapefile that represents shoreline geomorphology. 17. Coastal overlap should be a multiple of cell size. This multiple is the maximum number of pixels between two non-overlapping shorelines the model can cope with when processing geomorphology. If the geomorphology comes from local data and the shoreline is from a global dataset, the coastlines might not overlap completely. To resolve the discrepancy, increase coastal overlap to several multiples of cell size, otherwise, leave it at zero. 18. Specify the Natural Habitat directory (optional). The model can use optional polygon shapefiles that represent the location of various habitats. 19. Specify the Natural Habitat CSV table (optional). If the above input for natural habitat directory is specified, the model requires this table of habitat ranks and protective distance stored in a CSV. See the Data needs section for more information on creating and formatting this table. A sample CSV will is supplied. 20. Specify the climatic forcing grid (Wind-Wave data point shapefile). The model requires wind and wave statistics to create the wind and wave exposure variables. See the Data needs section for details on preparing a shapefile from another data source. 21. Specify the continental shelf layer (optional). To represent surge potential, the model uses a continental shelf polygon shapefile. 22. Specify the Sea Level Rise layer (optional). The model can use an optional polygon shapefile that represents sea level rise potential. 23. Specify the structures layer. Polygon shapefile that contains the location of rigid structures along or near the coast. 24. Specify the population layer. This file should be a raster population assigned to each cell value. The default data for this layer is a global raster located in InVEST/Base_Data/Marine/Population/global_pop. If users have a superior raster, they are instructed to select the location of this data on their local computer.
6.6. Running the model 73 InVEST User Guide, Release 2.5.6
25. Set the minimum population threshold in urban centers (default is 5000 people). The model will sum the population within a user-defined radius (see coastal neighborhood below) and will report segments that exceed the threshold. 26. Set the coastal neighborhood (default is 2000 meters). The model will sum the population within the specified radius. It is important to keep in mind that the surface over which the population is aggregated increases as the square of the radius. 27. At this point the model dialog box is completed for a complete run (with all optional data for full exposure analysis) of the Coastal Vulnerability model. Click the “run” button at the lower right corner of the window to run the model. A new window will appear and show the progress about each step in the analysis. It will also show the most salient warnings when preprocessing the input, as well as warnings during the computation of the various indices. Once the model finishes, the progress window will show all the completed steps and the amount of time that has elapsed during the model run. Upon successful completion of the model, two new directories called “intermediate” and “outputs” will be created in the workspace. The main outputs that are useful for further analysis are in the “coastal_vulnerability” and “population” sub-directories in “outputs”. The remainder of this guide will concentrate on these outputs. The types of spatial data that is generated are described in the Interpreting results section.
6.7 Interpreting results
6.7.1 Model outputs
The following is a short description of each of the outputs from the Coastal Vulnerability model. Files are grouped in sub-directories within the “intermediate/” and “outputs/” directories, except for “run_summary.html” that is directly located in “outputs/”. The “outputs/” and “intermediate/” directories are saved in the workspace directory that was specified by the user. Every sub-directory has a comma separated file (CSV) that is a text version of the compiled data of all raster files present in the subdirectory for each shoreline segment. Depending on the first option in the model’s user interface, these shore segments either cover the sheltered areas only, or both the sheltered and exposed ones.
Output directory
• outputs\run_summary.html – This file summarizes the run by showing four main pieces of information:
* A map of the area of interest, along with the AOI’s latitude and longitude. * A histogram of the vulnerability index (between 1 and 5) of the coastal segment * A histogram of the vulnerability of the population living along the coast * A histogram of the vulnerability of the urban centers along the coast. – outputs\coastal_vulnerability: contains all the layers used to compute the coastal vulnerability index.
* 1_a_shore_exposure.tif - a raster where the cells corresponding to the shoreline segments are either 0 if sheltered or 1 if exposed.
* 1_b_geomorphology.tif - a raster where shore segments are valued from 1 to 5 depending on the geomorphology in the geomorphology layer. Lower coastal values indicate geomorphologic types are less susceptible to erosion, and vice-versa.
* 1_c_relief.tif - a raster where shore segments are valued from 1 to 5 depending on the average elevation around that cell. Lower values indicate lower elevations.
6.7. Interpreting results 74 InVEST User Guide, Release 2.5.6
* 1_d_natural_habitats.tif - a raster where shore segments are valued according to the natural habitats that are present there. The model uses equation (6.4) that uses natural habitat ranks specified in Table 4.1.
* 1_e_wind_exposure.tif - a raster where the shore segments are ranked in equal proportion between 1 and 5 according to their exposure to winds.
* 1_f_wave_exposure.tif - a raster where shore segments are ranked in a similar way to wind exposure, but according to their exposure to wave.
* 1_g_surge_potential.tif - a raster where segments are ranked according to their exposure to potential surge. First, the exposed segments are assigned a rank in equal proportion between 1 and 5, depending on their distance to the edge of the continental shelf. Then, these values are propagated along the sheltered coast. Isolated coastline segments (such as islands) are assigned the rank of the closest (already ranked) segment.
* 1_h_sea_level_rise.tif - a raster with segments ranked in equal proportion between 1 and 5 based on the sea level rise value from the input shapefile.
* 1_i_coastal_vulnerability.tif - a raster with he coastal vulnerability index computed as in (6.1). * 1_j_coastal_vulnerability_no_habitats.tif - raster containing values computed from the same equation as the coastal vulnerability raster except the natural habitats layer has been replaced by the constant 5.
* 1_k_habitat_role.tif - raster difference between coastal_vulnerability_no_habitats and coastal_vulnerability.
* 3_2_erodible_shoreline.tif - raster where the shoreline segment values are computed with equation (6.3).
* coastal_vulnerability.csv - comma-separated file that aggregates the data in each file in the directory for each coastal segment – outputs\population: contains all the layers used to compute the coastal vulnerability index.
* 0_structures_edges.tif - a raster with only the shore segments that border the coastal structures. * 1_a_shore_exposure.tif - same as in the “coastal_vulnerability/” sub-directory. * 1_i_coastal_vulnerability.tif - same as in the “coastal_vulnerability/” sub-directory. * 1_j_coastal_vulnerability_no_habitats.tif - same as in the “coastal_vulnerability/” subdirectory. * 1_m_coastal_population.tif - raster where every coastal segment having the population living on the coast.
Intermediate directory
The model currently generates several hundreds of files classified in sub-directories in the intermediate directory. There is one intermediate subdirectory per computational step required to produce a file in the “outpus/” directory. Each of these sub-directory is prefixed so that the alphabetical order reflects the order of the model’s computational steps. Within a sub-directory, each file is the result of a computational step and is usually numbered so that it is possible to follow the order the computation that is carried out. If there is a problem with an output file, the user can go back to the corresponding sub-directory and look at the intermediate files individually to infer what happened during the computation. The explanation of each single file is going to be released in a later version of the model.
6.7. Interpreting results 75 InVEST User Guide, Release 2.5.6
6.7.2 Parameter log
Each time the module is run a text file will appear in the workspace directory. The file will list the parameter values for that run and be named according to the service and the date and time.
6.8 Appendix A
In this appendix, definitions for the terms presented in the geomorphic classification in Table 4.1 are presented. Some of these are from Gornitz et al. (1997) and USACE (2002). Photos of some of the geomorphic classes that are presented can be found at the National Oceanic and Atmospheric Administration’s Ocean Service Office of Response and Restoration website. Alluvial Plain A plain bordering a river, formed by the deposition of material eroded from areas of higher elevation. Barrier Beach Narrow strip of beach with a single ridge and often foredunes. In its most general sense, a barrier refers to accumulations of sand or gravel lying above high tide along a coast. It may be partially or fully detached from the mainland. Beach A beach is generally made up of sand, cobbles, or boulders and is defined as the portion of the coastal area that is directly affected by wave action and that is terminated inland by a sea cliff, a dune field, or the presence of permanent vegetation. Bluff A high, steep backshore or cliff Cliffed Coasts Coasts with cliffs and other abrupt changes in slope at the ocean-land interface. Cliffs indicate marine erosion and imply that the sediment supply of the given coastal segment is low. The cliff’s height depends upon the topography of the hinterland, lithology of the area, and climate. Delta Accumulations of fine-grained sedimentary deposits at the mouth of a river. The sediment is accumulating faster than wave erosion and subsidence can remove it. These are associated with mud flats and salt marshes. Estuary Coast The tidal mouth of a river or submerged river valley. Often defined to include any semi-enclosed coastal body of water diluted by freshwater, thus includes most bays. The estuaries are subjected to tidal influ- ences with sedimentation rates and tidal ranges such that deltaic accumulations are absent. Also, estuaries are associated with relatively low-lying hinterlands, mud flats, and salt marshes. Fiard Glacially eroded inlet located on low-lying rocky coasts (other terms used include sea inlets, fjard, and firth). Fjord A narrow, deep, steep-walled inlet of the sea, usually formed by the entrance of the sea into a deep glacial trough. Glacial Drift A collective term which includes a wide range of sediments deposited during the ice age by glaciers, melt-water streams and wind action. Indented Coast Rocky coast with headland and bays that is the result of differential erosion of rocks of different erodibility. Lagoon A shallow water body separated from the open sea by sand islands (e.g., barrier islands) or coral reefs. Mud Flat A level area of fine silt and clay along a shore alternately covered or uncovered by the tide or covered by shallow water.
6.9 Appendix B
The model requires large-scale geophysical, biological, atmospheric, and population data. Most of this information can be gathered from past surveys, meteorological and oceanographic devices, and default databases provided with the model. In this section, various sources for the different data layers that are required by the model are proposed, and
6.8. Appendix A 76 InVEST User Guide, Release 2.5.6 methods to fill out the input interface discussed in the Data needs section are described. It is recommend that users import all the required and optional data layers before attempting to run the model. Familiarity with data layers will facilitate the preparation of data inputs.
6.9.1 Population data
To assess the population residing near any segment of coastline, population data from the Global Rural-Urban Mapping Project (GRUMP) is used. This dataset contains global estimates of human populations in the year 2000 in 30 arc- second (1km) grid cells. User are encouraged to use their own, more detailed and/or recent census data, and it is encouraged that recent fine-scale population maps are used, even in paper form, to aid in the interpretation of the Exposure Index map.
6.9.2 Geo-physical data layer
To estimate the Exposure Index of the AOI, the model requires an outline of the coastal region. As mentioned in the Data needs Section, we provide a default global land mass polygon file. This default dataset, provided by the U.S. National Oceanic and Atmospheric Administration (NOAA) is named GSHHS, or a Global Self-consistent, Hierar- chical, High-resolution Shoreline (for more information, visit http://www.ngdc.noaa.gov/mgg/shorelines/gshhs.html). It should be sufficient to represent the outline of most coastal regions of the world. However, if this outline is not sufficient, we encourage that users substitute it with another layer. To compute the Geomorphology ranking, users must provide a geomorphology layer (Data needs Section, input 15) and an associated geomorphic classification map. This map should provide the location and type of geomorphic features that are located in the coastal area of interest. In some parts of the west-coast of the United States and Canada, such a map can be built from a database called Shorezone. For other parts of the United States, users can consult the Environmental Sensitivity Index website. If such a database is not available, it is recommend that a database from site surveys information, aerial photos, geologic maps, or satellites images (using Google or Bing Maps, for example) is built. State, county, or other local GIS departments may have these data, freely available, as well. In addition to the geomorphology layer, users must have a field in its attribute table called “RANK”. This is used by the model to assign a geomorphology exposure ranking based on the different geomorphic classes identified. Assign the exposure ranks based on the classification presented in Table 4.1. All ranks should be numeric from 1 to 5.
6.9.3 Habitat data layer
The natural habitat maps (inputs 7 and 8 in the Data needs Section) should provide information about the location and types of coastal habitats described in Table 4.1. The subtidal layers in that directory have been built from a database called Shorezone. Dune data from unpublished an dataset provided by Raincoast Applied Ecology was obtained. If such a database is not available, it is recommend building it from site surveys information, aerial photos, or even satellites images (using Google or Bing Maps, for example). The Natural Habitat CSV table input asks users to provide information about the type of habitats layers that users have in the “NaturalHabitat” directory. The different columns in that table are: 1. HABITAT: The name of the natural habitat for which users have a layer (e.g., kelp or eelgrass) 2. ID: The ID number associated with the name of these habitats: the underscored integer number X listed at the end of the name of the different layers that have been created, as in “eelgrass_2”. Note that this ID number is what the model uses to associate a rank and protection distance to the name users input in the first column. In other words, the name in column 1 can be different from the name of your file, but the ID number should match. For example, in the default natural habitat layers directory that has been provided, the eelgrass layer has the ID = 2 (e.g. eelgrass_2). Since the ID in the second column is 2, then the model recognizes that the rank and protection distance values that are defined for “eelgrass” apply to the eelgrass_2.shp layer.
6.9. Appendix B 77 InVEST User Guide, Release 2.5.6
3. RANK: The vulnerability rank associated with the natural habitat that is listed in column 1. It is recommend that the ranking system provided in Table 4.1 is used. However, if users would like to evaluate how the vulnerability index values changes in the absence of the habitats listed in the table, users should change the RANK to a 5. For example, to evaluate how the vulnerability of an area changes if high sand dunes are removed, users can change the RANK value for high sand dunes from a 2 to a 5. 4. PROTECTIVE DISTANCE (m): The model determines the presence or absence of various natural habitats that users specified in the AOI by estimating the fetch distance over the 16 equiangular segments between the location of the natural habitats and the shoreline. If there is a non-zero fetch distance between a patch of natural habitat and a shoreline segment, the model recognizes that the patch fronts that segment. To assign a natural habitat ranking to that segment which takes into account the beneficial effect of the presence of this habitat, it is asked that users input a maximum distance of influence into the Natural Habitat CSV table (input 8). It is assumed that natural habitats that are fronting a segment but are further away from the segment than the distance that is defined by the user will not have a beneficial effect on the stability of that segment, and will not be counted in the natural habitat ranking for that segment. To estimate this distance, it is recommend that users load the various habitat layers located in their “Natural Habitats” directory as well as the polygon layer representing the area of interest. Then, using the ArcGIS “distance” tool, measure the distance between the shoreline and natural habitats that you judge to be close enough to have an effect on nearshore coastal processes. It is best to take multiple measurements and develop a sense of an average acceptable distance that can serve as input. Please keep in mind that this distance is reflective of the local bathymetry conditions (a seagrass bed can extend for kilometers seaward in shallow nearshore regions), but also of the quality of the spatial referencing of the input layer. The example below gives an example of such measurement when seagrass beds are considered (green patches).
As mentioned in Natural Habitats, the model computes the natural habitat exposure ranking for a shoreline segment using the following equation: v u N u N 2 X 2 N 2 RHab = 4.8 − 0.5t(1.5 max(5 − Rk)) + ( (5 − Rk) − max(5 − Rk)) ) k=1 k=1 k=1
This equation is applied to various possible combinations of natural habitats, and the results of this exercise are presented in the table and figure below:
6.9. Appendix B 78 InVEST User Guide, Release 2.5.6
6.9. Appendix B 79 InVEST User Guide, Release 2.5.6
6.9.4 Wind data
To estimate the importance of wind exposure and wind-generated waves, wind statistics measured in the vicinity of the AOI are required. From at least 5 years of data, the model requires the average in each of the 16 equiangular sectors (0deg, 22.5deg, etc.) of the wind speeds in the 90th percentile or greater observed near the segment of interest to compute the REI. In other words, for computation of the REI, sort wind speed time series in descending order, and take the highest 10% values, and associated direction. Sort this sub-series by direction: all wind speeds that have a direction centered around each of the 16 equiangular sectors are assigned to that sector. Then take the average of the wind speeds in each sector. If there is no record of time series in a particular sector because only weak winds blow from that direction, then average wind speed in that sector is assigned a value of zero (0). Please note that, in the model, wind direction is the direction winds are blowing FROM, and not TOWARDS. For the computation of wave power from wind and fetch characteristics, the model requires the average of the wind speeds greater than or equal to the 90th percentile observed in each of the 16 equiangular sectors (0deg, 22.5deg, etc.). In other words, for computation of wave power from fetch and wind, sort the time series of observed wind speed by direction: all wind speeds that have a direction centered on each of the 16 equiangular sectors are assigned to that sector. Then, for each sector, take the average of the highest 10% observed values. Again, please note that, in our model, wind direction is the direction winds are blowing FROM, and not TOWARDS. If users would like to provide their own wind and wave statistics, instead of relying on WW3 data, the must enter the data in the following order: 1. Column 1-2: Placeholder. No information required. 2. Columns 3-4: LAT, LONG values. These values indicate the latitude and longitude of the grid points that will be used to assign wind and wave information to the different shoreline segments. 3. Columns 5-20: REI_VX, where X=[0,22,45,67,90,112,135,157,180,202,225,247,270,292,315,337] (e.g., REI_V0). These wind speed values are computed to estimate the REI of each shoreline segment. These values
6.9. Appendix B 80 InVEST User Guide, Release 2.5.6
are the average of the highest 10% wind speeds that were allocated to the 16 equiangular sectors centered on the angles listed above. 4. Columns 21 to 36: REI_PCTX, where X has the same values as listed above. These 16 percent values (which sum to 1 when added together) correspond to the proportion of the highest 10% wind speeds which are centered on the main sector direction X listed above. 5. Column 37 to 52: WavP_X, where X has the same values as listed above. These variables are used to estimate wave exposure for sites that are directly exposed to the open ocean. They were computed from WW3 data by first estimating the wave power for all waves in the record, then splitting these wave power values into the 16 fetch sectors defined earlier. For each sector, we then computed WavP by taking the average of the top 10% values (see Section The model). 6. Column 53 to 68: WavPPCTX, where X has the same values as listed above. These variables are used in combination with WavP_X to estimate wave exposure for sites that are directly exposed to the open ocean. They correspond to the proportion of the highest 10% wave power values which are centered on the main sector direction X (see Section The model). 7. Columns 69 to 84: V10PCT_X, where X has the same values as listed above. These variables are used to estimate wave power from fetch. They correspond to the average of the highest 10% wind speeds that are centered on the main sector direction X. If users decide to create a similar layer, it is recommend that they create it in Microsoft Excel, and add the sheet in the “Layer” menu. To plot the data, right-click on the sheet name, and choose “Display XY Data”. Choose to display the X and Y fields as “LONG” and “LAT”, respectively. If users are satisfied with the result, right-click on the layer, choose “Export Data” and convert this temporary “Events Layer” into a point shapefile that can now be called when running the Coastal Vulnerability model. Finally, make sure it has a WGS84 datum. As described in The model section Wind Exposure, the model provides an optional map of areas that are exposed or sheltered. This is purely based on fetch distances, and does not take into account measurements of wind speeds. To prepare this map, the model uses an estimate of a fetch distance cutoff to use that the user has defined, based on the AOI under consideration. To provide that distance, it is recommend that the “distance tool” on the global polygon layer, zoomed into the AOI, is used to determine that distance.
6.9.5 Sea level change
As mentioned earlier, a map of net rates of sea level rise or decrease in the AOI can be added. Such information can be found in reports or publications on Sea Level Change or Sea Level Rise in the region of interest. Otherwise, it is suggested that users generate such information from tide gage measurements, or based on values obtained for nearby regions that are assumed to behave in a similar way. A good global source of data for tide gauge measurements to be used in the context of sea level rise is the Permanent Service for Sea Level. This site has corrected, and sometimes uncorrected, data on sea-level variation for many locations around the world. From the tide gage measurements provided by this website, it is suggested that users estimate the rate of sea level variation by fitting these observations to a linear regression, as shown in the figure below. This figure was extracted from Bornhold (2008).
Create a sea level change GIS layer
Users can create their own polygon or points to represent the sea level change input to the model. If a polygon feature class is created, the model will apply the sea level change ranking assigned to that polygon for the segments of the shoreline that the polygon overlaps. If a point feature class is used, the model will assign sea level change rankings to shoreline segments based on whichever point is closest to that segment. To create a feature class in ArcMap, the map window must be in “data view” mode. Select the “Drawing” drop-down option and begin creating a polygon similar to the black feature below. Double click to complete the polygon. Similarly, you can select to create a 93marker94 rather than a 93polygon94 in the drawing tool bar. Select this option and click in locations throughout your area of interest where you would like to assign sea level change values or rankings. Next, click “Drawing >> Convert Graphics to
6.9. Appendix B 81 InVEST User Guide, Release 2.5.6
Features...” Specify the path of the output shapefile or feature class and a name that will clearly designate the extent. Finally, check the box: “Automatically delete graphics after conversion” and click “OK”. Once all polygons or points for specific regions are created, you must create an attribute field called “Trend” and populate it with values indicating net sea level change in mm/year according to Table 4.1. For more information on how to create a Sea Level Change layer, see the FAQ.
6.9.6 Surge potential
Surge potential is estimated as the distance between a shoreline segment and the edge of the continental shelf, or any other depth contour of interest. This output is computed using a method that does not take into account the presence of land barriers between a shoreline segment and the depth contour. When creating an AOI, loading the global polygon layer and the continental shelf (or other preferred depth contour, input 11) as guides is recommend. Draw the AOI so that it overlaps the portion of coastline you want to include in your analysis. Additionally, if you want to include the surge potential variable make sure the AOI overlaps at least a portion of the shelf’s closest edge to the coastline. This is necessary so that the model can properly calculate the distance to shelf.
6.10 References
Bornhold, B.D., 2008, Projected sea level changes for British Columbia in the 21st century, report for the BC Ministry of Environment. Bush, D.M.; Neal, W.J.; Young, R.S., and Pilkey, O.H. (1999). Utilization of geoindicators for rapid assessment of coastal-hazard risk and mitigation. Oc. and Coast. Manag., 42. Center for International Earth Science Information Network (CIESIN), Columbia University; and Centro Interna- cional de Agricultura Tropical (CIAT) (2005). Gridded Population of the World Version 3 (GPWv3). Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University. Cooper J., and McLaughlin S. (1998). Contemporary multidisciplinary approaches to coastal classification and envi- ronmental risk analysis. J. Coastal Res. 14(2):512-524
6.10. References 82 InVEST User Guide, Release 2.5.6
Finlayson, D. 2005, fetch program, USGS. Accessed February 2010, from http://sites.google.com/site/davidpfinlayson/Home/programming/fetch Gornitz, V. (1990). Vulnerability of the east coast, U.S.A. to future sea level rise. JCR, 9. Gornitz, V. M., Beaty, T.W., and R.C. Daniels (1997). A coastal hazards database for the U.S. West Coast. ORNL/CDIAC-81, NDP-043C: Oak Ridge National Laboratory, Oak Ridge, Tennessee. Hammar-Klose and Thieler, E.R. (2001). Coastal Vulnerability to Sea-Level Rise: A Preliminary Database for the U.S. Atlantic, Pacific, and Gulf of Mexico Coasts. U.S. Geological Survey, Digital Data Series DDS-68, 1 CD-ROM Irish, J.L., and Resio, D.T., “A hydrodynamics-based surge scale for hurricanes,” Ocean Eng., Vol. 37(1), 69-81, 2010. Keddy, P. A. (1982). Quantifying within-lake gradients of wave energy: Interrelationships of wave energy, substrate particle size, and shoreline plants in Axe Lake, Ontario. Aquatic Botany 14, 41-58. Short AD, Hesp PA (1982). Wave, beach and dune interactions in south eastern Australia. Mar Geol 48:259-284 Tolman, H.L. (2009). User manual and system documentation of WAVEWATCH III version 3.14, Technical Note, U. S. Department of Commerce Nat. Oceanic and Atmosph. Admin., Nat. Weather Service, Nat. Centers for Environmental Pred., Camp Springs, MD. U.S. Army Corps of Engineers (USACE). 2002. U.S. Army Corps of Engineers Coastal Engineering Manual (CEM) EM 1110-2-1100 Vicksburg, Mississippi. Wessel, P., and W. H. F. Smith (1996). A Global Self-consistent, Hierarchical, High-resolution Shoreline Database, J. Geophys. Res., 101, #B4, pp. 8741-8743. World Resources Institute (WRI) (2009). “Value of Coral Reefs & Mangroves in the Caribbean, Economic Valuation Methodology V3.0”. Show details
6.10. References 83 CHAPTER SEVEN
EROSION PROTECTION MODEL
7.1 Summary
Understanding the role that nearshore habitats play in the protection of coastal communities is increasingly impor- tant in the face of a changing climate and growing development pressure. The InVEST Erosion Protection model quantifies the protective benefits that natural habitats provide against erosion and inundation (flooding) in nearshore environments. It is composed of two sub-models: a Profile Generator and a Nearshore Wave and Erosion model. In the absence of local data detailing the profile of the nearshore elevations, the Profile Generator model helps you combine information about the local bathymetry and backshore to generate a 1-Dimensional (1D) cross-shore (perpendicular to the shoreline) beach profile. The Nearshore Waves and Erosion model uses the cross-shore profile (either uploaded or created using the Profile Generator) to compute summaries of nearshore wave information, and outputs the total water level at the shore, the amount of shoreline erosion, and the amount of avoided damages due to erosion (in your local currency) from a given habitat management decision that affects the amount of nearshore marine habitats (e.g., coral or oyster reefs, vegetation, sand dunes) at your site. This information can help coastal managers, planners, landowners and other stakeholders understand the coastal protection services provided by nearshore habitats, which can in turn inform coastal development strategies and permitting. The model, of course, has some limitations (see Limitations and Simplifications), however, all the science that went into this model is based on well-established models that have been developed and successfully tested by various scientists at many sites, and is expected to be useful for a wide range of management decisions. Together with the “Tier 0” Coastal Vulnerability model, this “Tier 1” model makes up InVEST’s Coastal Protection toolbox.
7.2 Introduction
The Erosion Protection model estimates, by way of a 1D bathymetry transect (or a series of transects) perpendicular to the shoreline, the relative contributions of different natural habitats in reducing nearshore wave energy levels and coastal erosion. These habitats dissipate wave energy and/or act as barriers against high waves and high water levels, and eventually protect coastal properties and communities. The protective services offered by those habitats to coastal populations are highlighted by linking coastal storms to erosion, and then linking erosion to damage of coastal prop- erties. To measure these services, the model has you specify a baseline habitat and a proposed scenario. The proposed scenario can incorporate habitat losses, gains, or some combination of the two over the area of interest. The service provided by these habitats can be measured by the amount of avoided erosion or inundation, by the number of people protected, or by the value of avoided property damages. The Erosion Protection model is composed of two sub-models: a Profile Generator and a Nearshore Waves and Erosion model. The purpose of the Profile Generator is to assist in the preparation of a 1D bathymetry transect for use in the Nearshore Waves and Erosion model. The inputs of the Profile Generator include the site’s location, a file describing the overall shape of the shoreline, and a nearshore topography and bathymetric file that contains land elevation as well as water depths near the site of interest. Furthermore, the model requires information about sediment size, tidal range and backshore characteristics, which is to be input in the Erosion Protection Excel Table (see Erosion Protection
84 InVEST User Guide, Release 2.5.6
Excel Table). If this information is not available, we provide guidance on how to approximate these inputs. Outputs of the Profile Generator model include a 1D bathymetry profile at your specified location, information about the site’s backshore and the location of natural habitats along the cross-shore transect from offshore to the uplands. In addition, the model provides over-water fetch distances (the distances over which wind blows over water to generate waves) as well as estimates of wave height and wind speeds that can occur at the site of interest during a storm. Overall, this model generates the inputs you need to run the Nearshore Waves and Erosion model. The Nearshore Waves and Erosion model uses information about the type and location of natural habitat at your site, as well as the way in which your scenario changes those habitat characteristics, to produce estimates of the amount of avoided shoreline erosion or scour and the associated change in property damages. The model inputs are a 1D bathymetry profile (obtained from the Profile Generator model or a site survey) and a value for offshore wave height and period (or a value of wind speed, fetch and the average water depth) that are representative of a certain storm conditions that also have to be described. In addition, the model requires information about the backshore as well as the type and physical characteristics of the natural habitats that are at the site of interest. You will need to specify how your scenario, or management action, will affect the natural habitats footprint or change their density, as compared to the baseline conditions. Model outputs are profiles of wave height under baseline and proposed scenario conditions. The model also estimates the amount of avoided shoreline retreat of a sandy beach or the volume of sediment eroded from consolidated beds (e.g., scour of mud bed) due to your management actions. When you select the valuation option, the model uses the biophysical outputs of the Nearshore Wave and Erosion model coupled with your input about local property value to provide you with information about the change in property damage associated with your scenario. With this model you can answer a variety of questions. The most basic answer it provides is the incremental change in erosion and property damages from erosion due to a habitat management action. This is simply the net differences in damages between the baseline and proposed scenario. You can also use this model to answer questions about the total level of erosion reduction service provision by habitats in an area. This could be approximated linearly from the incremental value derived from your habitat management action model run. Alternatively, you could propose a scenario where no service provision is provided by habitat and estimate the total value of habitats in that way. Refer to the limitations and simplifications section below for more information on how to interpret these results. This model should be thought of as estimating one part of the environmental services associated to habitats: protection from erosion. When conducting a cost-benefit analysis, it is important to evaluate all value changes arising from a proposed management action. We recommend you run the Tier 1 model after the Coastal Vulnerability model, which is also part of the “Coastal Protection” tool box. The Tier 0 Coastal Vulnerability model maps regions that are more or less vulnerable to erosion and inundation during storms and also highlights important characteristics of the region of interest. In addition, it maps regions of the shoreline that are exposed to or sheltered from the open ocean and estimates wind-generated wave characteristics. Coastal Vulnerability model outputs that explore the effects of various management actions (e.g., the presence vs. absence of natural habitats) will help identify regions where natural habitats or a certain management action may have significant impacts on the stability of the coastline. However, the Coastal Vulnerability (Tier 0) and Nearshore Wave (Tier 1) models are independent. You do not need to run the Coastal Vulnerability model in order to run the Erosion Protection model.
7.3 The model
The InVEST Erosion Protection model is a 1D process-based tool that produces an estimate of wave attenuation and erosion reduction owing to the presence of natural habitats, as well as an estimate of the value of those habitats in your local currency unit. A single model run operates along a single transect perpendicular to the shoreline; multiple runs can be distributed along broader swaths of coastline to explore the protective services of natural habitats and the effects of various management actions on the hazards of erosion and flooding within larger regions. At this point, however, the model does not batch process multiple runs so each transect is run on an individual basis.
7.3. The model 85 InVEST User Guide, Release 2.5.6
7.3.1 How it works
As waves travel from the deep ocean to coastal regions with shallower waters, they start to interact with the seabed. They first increase in height before breaking and dissipate most of their energy in the surf zone and on the beach face. Natural habitats play an important role in protecting shorelines against wave action because they increase the amount of wave dissipation, or, in the case of sand dunes, serve as a physical barrier. To estimate the profile of wave height that one would expect at a certain region as the wave propagates shoreward three types of information are required: 1. Offshore wave characteristics: wave height and wave period at the deepest point in the bathymetry profile. 2. Nearshore bathymetry and backshore characteristics: elevation relative to Mean Lower Low Water (MLLW) of both the submerged (underwater) and emerged (above water) portions of the cross-shore profile. 3. Location and physical characteristics of natural habitats: distance from the shoreline of the natural habitats that will become submerged during a storm, as well as representative density, height and diameter of the habitat elements. The InVEST Erosion Protection model is composed of two sub-models. The first model, the Profile Generator, helps you obtain cross-shore transect containing nearshore bathymetry and topography information at your site. The Profile Generator can also place the footprints of the natural habitats along that transect. Using this cross-shore profile (or one that you upload), the Nearshore Waves and Erosion model computes profiles of wave height and wave-induced mean water level for your baseline and post-management action scenario. This scenario generally affects the footprint and/or density of the various habitats that you can have at your site. Currently, the model is suitable to value the following sub-tidal (always submerged), inter-tidal (between high and low tides) and supra-tidal (above the high-water mark) habitats: seagrass beds, marshes, mangroves or coastal forests, coral reefs and oyster reefs. The remainder of this section will describe how both the Profile Generator and the Nearshore Waves and Erosion models work.
Profile Generator Model
In order to run the Nearshore Wave and Erosion model, it is necessary to have nearshore bathymetry (water depths) and topography (land elevation) information, as well as the location and characteristics of natural habitats at your site. Also, you must provide an offshore wave height and associated period values. The purpose of the Profile Generator model is to help you glean this information from your site data and help you prepare to run the Nearshore Wave and Erosion model. Additionally, the Profile Generator helps you estimate those data (nearshore elevations and slopes) if you do not have them but know the general characteristics of the site. First, the Profile Generator helps you obtain bathymetry information by three different options. The model inter- face asks: “Do you want us to cut a cross-shore transect in GIS?”. If you have a seamless topography/bathymetry (topo/bathy) Digital Elevation Model (DEM) or a bathymetric DEM, you should answer “Yes”, and upload a Digital Elevation Model (DEM). If the uploaded DEM is a seamless DEM with both bathymetry and topography represented, the Profile Generator will capture both topographic as well as bathymetric information. Otherwise, if the DEM only captures bathymetric elevations and excludes land elevations above the water level, the cut profile will apply erroneous values of 100.0 (meters) for the missing terrestrial portions. You will have the opportunity to create an ideal backshore profile by filling appropriate information in the Erosion Protection Excel Table (see below). When you choose the option “Do you want us to cut a cross-shore transect in GIS?”, the Profile Generator works by drawing a transect perpendicular to the shoreline of your site of interest, of a length that you specify in the model interface. This length should be such that a sufficiently deep point is reached and that any adjacent land features are excluded. The model reads the bathymetric and, if the DEM provided is seamless, topographic information along that transect. If the site is surrounded by land (sheltered), or is fronted by an island, the offshore portion of the profile might include the adjacent land feature where waves cannot propagate. To avoid this situation, the model removes any portions of the profile offshore of the deepest point that is shallower than the average depth along the profile. Another option is to answer “No, but I will upload a cross-shore profile”, and then to upload a profile obtained from another source for further processing. At the very least, two coordinate points (X-cross-shore distance from shoreline,
7.3. The model 86 InVEST User Guide, Release 2.5.6
and Z-elevation relative to Mean Lower Low Water [MLLW]) in the uploaded file are required. Lastly, if you do not have bathymetric information at the site of interest, you can choose the third option “No, please create a theoretical profile for me”, and the model will generate, for sandy systems only, a theoretical bathymetric profile, based on the average sand size at the site. The depth profile follows the equation (Dean and Dalrymple, 2002, Chap. 7):
Z = AX2/3 (7.1) where (X,Z) represent the cross-shore distance and depth, with X = 0 at the shoreline. The coefficient A is a profile scale factor and is a function of sediment size (Dean and Dalrymple, p.162 and CEM). This shape of the bed profile is called an equilibrium beach profile, and corresponds to the average profile that one would obtain after averaging years of regular bathymetric surveys at a sandy beach. It can also be viewed as a profile that develops when destructive and constructive forces are in equilibrium. Usually, this profile extends to what’s called the “closure depth”, which is the depth where waves no longer affect sediment movement on the bottom. However, for simplicity, this profile is extended from the water line down to -20 meters. Please remember that this option is only valid for sandy systems, where sediment size varies between 0.1 to 1.09 mm. Further it is most applicable to oceanic or exposed shorelines (recall, T0: Coastal Vulnerability helps designate exposed versus sheltered coastlines). Once the method that will be used to create an initial bathymetry profile is selected, the Profile Generator will help you modify or add to the information contained in that transect in order to represent the site as accurately as possible. This is especially useful if you want to estimate the amount of erosion at your site and important backshore details are not captured in the elevation (DEM) or habitat (Polygon Shapefiles) inputs. The Profile Generator will also help you modify, smooth or remove portions of the profile to represent the effects of a management action under consideration (e.g., remove offshore portions that are too deep to affect wave heights or remove certain bathymetric features before conducting the analysis). In order to best use the Profile Generator, it is important that you are familiar with some terminology and typical val- ues of backshore slopes for different environments. Figure 1 shows profiles of a typical beach and a coastal mangrove forest. After waves have propagated from deep water and broken in the nearshore, they reach the foreshore and/or backshore portion of the beach, or, if the water level is high enough, propagate through a marsh or a mangrove forest. Under normal conditions, for sandy beaches there is a relatively flat region between the Mean Lower Low and Mean Higher High (MHHW) water marks called the foreshore. The backshore (the region above MHHW) consists of a berm and, in temperate regions mostly, a sand dune. Berms can range in width from 10’s of meters to having a very small or no width. In general, foreshore and backshore information cannot be obtained during standard hydrographic surveys. Also we have found that, although most DEM files have relatively good bathymetric information, intertidal and backshore elevations are often incorrect, unless they were measured during a detailed topographic survey effort. Mangrove forests are usually fronted by a tidal flat with an average slope of 1:1000 to 1:2000, and usually have a rela- tively monotonic profile whose slope varies between 1:200 to 1:600 (de Vos, 2004; Burger, 2005). In case you would like to measure foreshore and backshore profiles at your site, you can either use standard surveying methods, or follow the simple method in Appendix A. However, if you cannot conduct such a survey, you can use the recommendations provided in this guide. In the Erosion Protection Excel Table, which summarizes the pertinent characteristics of the profile, you must indicate whether the profile of interest is a sandy beach or a muddy backshore. This option determines what modifications may be made to the cut or user-defined topo/bathy profile. Option 1. Add backshore to a sandy beach: assuming that this information is not contained in the cross-shore profile that was cut by the model or in the profile you uploaded, the Erosion Protection Excel Table (see Erosion Protection Excel Table) helps you guess what the foreshore slope, berm height and dune height might be for the site of interest, based on simple rules of thumb. Please bear in mind that conditions at the site of interest can differ quite substantially from these rules. Therefore, the suggestions provided should be used as a starting point but a site survey (even as basic as field notes from visual observations) is strongly encouraged if you are interested in obtaining more accurate results. The average sediment size is required to help approximate foreshore slopes, in case you do not have it. It also helps creating a bathymetric profile for sandy beaches in case you do not have any site measurement. Finally, it is used in the erosion model of the Nearshore Waves and Erosion model. As mentioned earlier, the foreshore is the intertidal region of the beach profile and is assumed to be linear in the Tier 1 model. To provide guidance on what that slope might be,
7.3. The model 87 InVEST User Guide, Release 2.5.6
Figure 7.1: Typical profiles of a sandy beach (top subplot) and a mangrove forest (bottom subplot). Please note the locations of the foreshore in the sandy beach as well as the slope of the mangrove forest.
five different values of slope, based on the sediment size, are provided. The first three are derived from observations presented in Wiegel (1964) at beaches that are protected, moderately exposed or fully exposed to the open ocean, in the U.S. The fourth value is derived from observations by McLachlan and Dorvlo (2005) at various beaches around the world. The fifth value is the average of the four previous values. If you do not have the precise sediment size, you can select a value based on a qualitative description of the sand (very fine, fine, medium, course, or very course). Berm height and foreshore slope often change as a function of seasonal wave climate. After a storm, the profile is flatter and the berm is lower than during fair weather conditions. However, in case you do not have any information about berm height at the site, it is recommended that you place the berm at least at 1 meter above the MHW mark. Finally, a dune height value is needed. Dunes are fairly common in temperate climates and height estimates can be derived from site surveys. In case you do not have this information, we recommend that you enter a value of 2 meters in order to get an estimate of how dunes can protect your site. Option 2. Add a backshore to a mangrove or marsh. Mangrove and marsh beds are different from sandy beaches because they consist, in general, of consolidated materials, do not have dunes, and their profile is fairly linear. As mentioned earlier and shown in Figure 1, mangrove forests are usually fronted by a tidal flat with an average slope of 1:1000 to 1:2000, and have a relatively monotonic profile whose slope varies between 1:200 to 1:600 (de Vos, 2004; Burger, 2005). If this option is selected, you can enter a maximum of three linear slopes that can be added to the bathymetry profile that you cut/created or uploaded. The Profile Generator locates the presence of natural habitats along the cross-shore profile. If Option 1 “Yes” to the question “Do you want us to cut a cross-shore transect in GIS?” is selected, you can also indicate the types of natural habitats that are present in the region of interest, and the model will locate and plot where those habitats fall onto the cross-shore transect. This is done by providing the path to the directory containing separate polygon shapefiles representing the footprints of each habitat. Please note that the results for the habitat placement are accurate only if the natural habitat and bathymetry layers are properly geo-referenced. You should scrutinize results to make sure that the natural habitats are properly placed along the profile (e.g., make sure that seagrass beds are in subtidal areas, or mangroves are in inter- or supra-tidal areas). Finally, if you do not have any storm wave or wind information at the site to run the Nearshore Waves and Erosion model, the Profile Generator will help you obtain those data by reading and providing you with some pre-processed statistics from the closest WAVEWATCH III (WW3, Tolman (2009)) grid point. Because wave data can be scarce
7.3. The model 88 InVEST User Guide, Release 2.5.6
in most regions of the world, 7 years of WW3 model hindcast reanalysis results have been analyzed to estimate, for model grid points that are in waters deeper than 50m, the maximum as well as the average of the top 10% and 25% wave height. The same statistics for wind data, for 16 equiangular direction sectors (0deg, 22.5deg, 45deg, etc.) have also been calculated. Wind information can be used in the Nearshore Waves and Erosion model by combining it with fetch distance (the distance over which waves are generated by wind) as well as the average depth offshore of the site to compute an offshore wave height and period. The model can compute these fetch distances if you choose Yes to the question Do you want the model to compute fetch distances?. In that case, from the site location, the model draws 16 equiangular sectors, and in each sector, the model draws nine equiangular radials. Each radial is initially 50km long, but is cutoff when it intersects with a land mass. To capture the effects of those land masses that limit fetch distance, the average fetch distance Fk for each 22.5deg sectors k is weighted by each radial distance and angle (Keddy, 1982):
P9 f cos θ F = n=1 n (7.2) k P9 n=1 cos θ
th th where fn is the n radial distance in the k equiangular sector, and θ = 2.5deg (22.5deg divided by 9). From wind speed, and fetch distance, wave height and period of the locally generated wind-waves are computed for each of the 16 equiangular sectors as:
h −4 0.79 i0.572 1.14 2.14.10 Fe H = He∞ tanh 0.343de tanh tanh(0.343de1.14) (7.3) h −7 1.45 i0.187 2.01 2.77.10 Fe T = Te∞ tanh 0.1de tanh tanh(0.1de2.01)
where the non-dimensional wave height and period He∞ and Te∞ are a function of the average wind speed values U 2 2 that were observed in a particular sector: He∞ = 0.24U /g, and Te∞ = 7.69U /g, and where the non-dimensional fetch and depth Fe∞ and de∞ are a function of the fetch distance in that sector Fk and the average water depth in the 2 2 2 region of interest d[m]: Fe∞ = gF/U , and Te∞ = gd/U . g[m/s ] is the acceleration of gravity. This expression of wave height and period assumes fetch-limited conditions (USACE, 2002; Part II Chap 2). Hence, model results may over-estimate wind-generated wave characteristics at a site if the duration of wind steadily blowing in a fetch direction is less than the time required to realize fetch-limited conditions. Also, wind-waves are not appropriate representations of wave climate on exposed, oceanic coasts. For oceanic coasts, estimates of representative oceanic wave forcing should be used (from WW3 data or another source) rather than wind-wave estimates. Once a satisfactory bathymetry and topography profile and realistic wave parameters are obtained, you can run the wave Nearshore Waves and Erosion model.
Nearshore Waves and Erosion
The amount of shoreline retreat at sandy beaches is a function of the total water level at the site and storm duration. Conversely, the erosion of muddy shorelines is a function of wave forces on the bed and storm duration. The total water level at the shoreline is composed of the sum of storm surge, wave runup, tide, amount of sea-level rise and any water surface elevation anomaly (e.g., super-elevation during an El Niño). To quantify the protective services provided by natural habitats, the Erosion Protection model computes the amount of attenuation of waves and the reduction in wave-induced mean water level increase (runup) at the shoreline caused by your scenario, as well as the difference in wave-induced velocity at the bed. This information is translated into an avoided erosion amount based on your scenario input, as well as a change in the protective service value of habitats, expressed in your local currency unit.
7.3. The model 89 InVEST User Guide, Release 2.5.6
Wave Evolution Model
The first step in this model is to estimate the waves that will “attack” the shoreline. Assuming that waves have a deep water height of Ho and a period T , it is possible to compute the evolution of wave height from offshore to the shoreline along the x-axis of cross-shore transect that you defined with the following wave energy equation:
1 ∂C H2 ρg g = −D (7.4) 8 ∂x where ρ is the density of seawater, taken as 1, 024kg/m3, g = 9.81m/s2 is the gravitational acceleration, H is the wave height representative of the random wave field, Cg is the speed at which wave energy travels, and D represents the dissipation of wave energy. The role of dissipation is to decrease the amount of wave energy as it propagates through or over different media. It is the sum of the dissipation caused by wave breaking DBreak, bottom friction DBot, and submerged vegetation DV eg :
D = DBreak + DV eg + DBot (7.5)
Dissipation due to breaking is modeled using the formulation and default parameters presented by Alsina and Baldock (2007), which performed well when compared to various field measurements, even without calibration (Apostos et al., 2008): " ! ! √ # H3 H 3 3H H 2 3 π H D = A b + b ) exp − b + 1 − erf b (7.6) Break h H 2H H 4 H
where erf is the Gauss error function, h is the local water depth, A is the sediment scale factor (see Profile Generator Model), and Hb is the maximum wave height prior to breaking: 0.88 kh H = tanh γ (7.7) b k 0.88 where k is the wavenumber, the ratio of length between two wave crests (called wavelength) L to 2π, and γ is a calibration parameter called the breaking index. The breaking index value, γ, used in the model is the value proposed by Battjes and Stive (1985):
H γ = 0.5 + 0.4 tanh 33 o (7.8) Lo
where Ho and Lo are the deepwater wave height and wavelength, respectively. The other dissipation terms in Equation (7.5) are expressed as a function of the characteristics of the natural habitats that are present along the profile of interest. In the model, as waves move into portions of the profile with natural habitat, this dissipation term is included. Any non-linear processes that might occur as waves move from one medium or habitat to another are ignored in the model. Dissipation due to the presence of vegetation is expressed by (Mendez and Losada, 2004):
3 3 1 kg sinh kαh + 3 sinh kαh 3 DV eg = √ ρNdCd H (7.9) 2 π 2σ 3k cosh3 kh
7.3. The model 90 InVEST User Guide, Release 2.5.6
where N is the density of vegetation (stems per unit area), d is the frontal width or diameter of vegetation stems, and hc α represents the fraction of the water depth h occupied by vegetation elements of average stem height hc: α = h . In the case of emergent vegetation (hc > h), a maximum of α = 1 is applied.
Finally, Cd is a taxa-specific (e.g., eelgrass, marsh, mangroves) drag coefficient. Default values of drag coefficient (see e.g., Kobayashi et al., 1983; Bradley and Houser, 2009; Burger, 2005 ) are applied in the model:
• For seagrass beds and marshes, Cd = 0.01
• For trees, including mangroves, Cd = 1 For trees, and mangroves in particular, we assumed that roots, trunk and canopy contribute independently to the total dissipation caused by vegetation, and DV eg becomes: DV eg = DRoots + DT runk + DCanopy. In addition to dissipation caused by vegetative elements, waves can also lose energy because they propagate over a rough bottom such as a coral reef top. Dissipation due to bottom friction is generally initiated when waves are in shallow enough water to “feel” the bottom, and is higher for coarser bed material than smoother ones. In the model, it is triggered when waves travel over sandy bottoms, as well as coral reefs, which are rougher than sand beds. Following Thornton and Guza (1983), the dissipation due to bottom friction is modeled as: 1 σH 3 D = ρC √ (7.10) Bot f 16 π sinh kh
where Cf is the bed friction coefficient, which is a function of the roughness (or dimensions) of the bed, and σ is the wave frequency, the ratio of wave period T to 2π. In the model, the following default friction coefficients have been assumed:
• For live corals, Cf = 0.2,
• For dead (smooth) corals that are still structurally stable : Cf = 0.1
• For corals that are structurally compromised and sandy beds: Cf = 0.001, The wave-evolution equation (Equation (7.4)) is valid when the bottom slope is not too steep. When waves encounter a steep barrier such as a coral reef, the model does not compute the amount of breaking dissipation and the profile of wave height during breaking. However, the value of the broken wave height at the edge of the reef top Hr is estimated assuming that wave height is controlled by water depth htop (Gourlay, 1996a, b) : Hr = 0.46htop, where htop = hr + ηr + h+ is the total water depth on top of the reef.
The total water depth is the sum of the depth on the reef top referenced to Mean Sea Level hr, the wave setup on the reef caused by breaking waves ηr, and any additional super-elevation of the water level η+, which can be caused by tides, pressure anomalies, etc. The wave setup on the reef top is caused by the release of wave energy during breaking and it is computed using the empirical equation proposed by Gourlay (1996a,b; 1997): √ 3 gH2T η = K i (7.11) r 64π p 3/2 (ηr + hr)
where Hi is the incident wave height, or the wave height at the offshore edge of the coral reef. The coefficient Kp is the reef profile shape factor, and is a function of the reef face slope αf or the reef rim slope αr, depending on whether waves break on the reef face or rim. Once the broken wave height is established following the equation presented above, the profile of wave height over the reef top is determined following Equation (7.4), with DBot as defined in Equation (7.10). Similar to coral reefs, when waves encounter a steep barrier such as an oyster reef, the amount of breaking dissipation is not computed. Instead, the model estimates the wave height Ht immediately shoreward of the reef with the following equations based on the incident wave height Hi immediately offshore of the reef:
Ht = KtHi (7.12)
7.3. The model 91 InVEST User Guide, Release 2.5.6
where Kt is a transmission coefficient. In the case of trapezoidal-shaped reefs, the transmission coefficient is computed with an empirical formula developed for low-crested breakwaters (van der Meer et al., 2005):
−0.31 Rc B −0.5ξ −0.4 + 0.64 1 − e if B/Hi < 8 Hi Hi Kt = −0.65 (7.13) Rc B −0.41ξ −0.35 + 0.51 1 − e if B/Hi > 12 Hi Hi
where B is the crest width of the reef, and Rc = hc − h is the crest freeboard, the difference between the reef height 0.5 hc and the water depth h. The breaker parameter ξ is computed as ξ = tan α/ (Si) where the seaward slope of the reef tan α is computed as a function of the structure crest and base width, B and W , respectively: 2h tan α = c (7.14) W − B
Finally, Si is the incident wave steepness:
2 Hi Si = (7.15) pi gTp
In the above equation, when 8 < B/Hi < 12, Kt is estimated by a linear approximation. If the oyster reef is a dome, the model applies empirical equation proposed by Armono and Hall (2003): H h h K = 1.616 − 4.292 i − 1.099 c + 0.265 (7.16) t T 2 h W
Once waves have travelled past the coral and oyster reefs, the evolution in the remaining portion of the bathymetry is modeled using the wave evolution equation (Equation (7.4)). It is assumed that the peak period T does not change.
Nearshore Bed Erosion
The next step is to model the response of the shoreline to wave attack. The model estimates two types of shoreline response. In sandy beach systems, the amount of shoreline retreat that takes place after a storm is approximated based on the value of storm surge that you input, and the value of wave runup computed by the wave evolution model. When the shoreline is composed of consolidated sediments (mangroves, marshes), the model estimates an hourly amount of bed scour and computes the volumetric sediment loss based on scour rate and storm duration. In both cases, empirical equations are used that ignore the dynamic feedback that takes place between wave and bed as the erosion occurs.
Wave runup (R2; see USACE (2002, Chap. 4)) is an estimate of the maximum shoreward distance that waves can reach on inundated lands. Once the profile of wave height has been computed, the amount of wave runup at the shoreline is estimated based on the empirical equation proposed by Stockdon et al. (2006): p p 2 R2 = 1.1 0.35m HoLo + 0.5 0.563m HoLo + 0.004HoLo (7.17)
where m is the foreshore slope, or the average cross-shore slope at the shoreline. In the above equation, the first term in the parenthesis represents the wave setup, and it can be influenced by the presence of the vegetation. The second term represents the wave swash, and it is composed of two terms. The first term, which is a factor of the foreshore slope m is called incident wave swash, and it can also be influenced by the presence of the vegetation. The second term is the called the infragravity swash. It is assumed that this term is not affected by the presence of vegetation
7.3. The model 92 InVEST User Guide, Release 2.5.6
elements because vegetation does not affect long-period waves as much as it does short period waves (Bradley and Houser, 2009). In the absence of biogenic features, the Erosion Protection model only requires information on the characteristics of offshore waves and foreshore slope to compute wave runup with Equation (7.17). If intertidal or subtidal biogenic features are present, wave runup is estimated via a series of steps described below. First, the wave height profile is estimated, in the absence and in the presence of vegetation, following the procedure outlined above. From these wave height profiles, the wave setup η at the shoreline is estimated by solving the following force balance equation: ∂S ∂η xx + ρg (h + η) − f = 0 (7.18) ∂x ∂x x
where Sxx is the force per unit length generated by the waves on the water column, and fx is the force per unit area due to the presence of vegetation elements:
fx = −αFx (7.19)
where the force Fx is computed following Dean and Bender (2006): 1 k F = ρg NdC H3 (7.20) x 12π d tanh kh
Neglecting non-linear processes associated with wave propagation, this equation is only valid for emergent vegetation. Consequently, the coefficient α is added to approximate the effects of vegetation on the wave setup when it is sub- merged. This approximation over-estimates the reduction in wave setup caused by submerged vegetation compared to what would be obtained if a non-linear wave theory to estimate Fx were adopted. However, this approximation is much faster and simpler to adopt. Once a value of wave setup in the absence of vegetation has been obtained, a proportionality coefficient β between the empirical estimate of wave setup and the value of the modeled wave setup at the shoreline ηShore is computed: η β = shore√ (7.21) 0.35m HoLo
v Based on the modeled value of the wave setup at the shoreline in the presence of vegetation, ηShore, the hypothetical offshore wave height Hp that would have achieved the same modeled setup is computed, assuming that the value of the coefficient β is the same:
v 2 1 ηShore Hp = (7.22) Lo 0.35m
v In cases when the effects of vegetation are so pronounced that ηShore is negative, it is assumed that Hp = 0. We adopted this empirical approach as a way to estimate the way in which vegetation affects runup, in the absence of observations or models.
Finally, to estimate the amount of runup at the shoreline in the presence of natural habitats, Ho is replaced in Equation (7.17) by the value of the hypothetical offshore wave height Hp in the wave setup and wave-induced swash terms: q p 2 R2 = 1.1 0.35m HpLo + 0.5 0.563m HpLo + 0.004HoLo (7.23)
7.3. The model 93 InVEST User Guide, Release 2.5.6
where the last term is left untouched because, as mentioned earlier, it has been assumed that long waves are not affected by the presence of natural habitats. Similarly, the value of the offshore wavelength Lo is not changed because it has been assumed that peak wave period is not affected by the presence of natural habitats. From the value of runup at the shoreline, the amount of beach retreat (sandy berm) or volumetric sediment loss (mud) can be computed. Sandy beaches are eroded during storms and generally build back during periods of fair weather. The amount of shoreline erosion is a function of the elevations of sand berm and dunes in the backshore, the wave height and period during the storm, the length of the storm and the total water level elevation during the storm. As mentioned earlier, the total water level during the storm is a function of the storm surge elevation, wave runup elevation, the tide stage during the storm and any super-elevation of the water surface caused by large-scale oceanic processes (e.g. El Nino). In the model, a storm surge elevation value is required as input and as well as offshore (starting) wave height and period. From these forcing inputs, the model computes the amount of runup for the different management actions that you wish to evaluate from Equation (7.17). Consequently, it is important that you adjust the bathymetry profile to any other water surface elevation difference that you wish to evaluate in the model. For example, if you are interested in investigating wave inundation and erosion at high tide, the elevation of high tide should be added to the value of the surge for a given storm.
The distance of sandy beach retreat during a storm Es is estimated following the model proposed by Kriebel and Dean (1993): 1 E = − (1 − cos α)E (7.24) s 2 ∞
where the beach potential erosion response if the storm lasted an infinite amount of time E∞ is scaled by the duration of the storm under consideration by a time-correction factor α. The potential erosion response E∞ is computed as a function of the wave breaking characteristics and the backshore dimensions:
S(xb − hb/m) − W (B + hb − 0.5S) E∞ = (7.25) B + D + hb − 0.5S where S is the total water level during the storm, referenced to MSL (please note that the model adjusts the bathymetry to MSL based on the tide information that you provided in the Erosion Protection Excel Table, so the initial bathymetry profile should be referenced to MLLW). hb and xb represent the water depth and distance from the shoreline where the offshore wave breaks with a height Hb. Breaking wave characteristics are computed by applying the wave evolution equation, Equation (7.4), to an equilibrium profile built from the sediment scale factor correspond- ing to the sediment size at the site (see Equation EqProf in Profile Generator Model). E∞ is also a function of the foreshore slope m, as well as the height and width of the sand berm B and W , and dune height D in the backshore, as well as, the specified berm height, B, and breaking depth, hb. Equation (7.25) is only valid up to a certain maximum surge elevation. E∞ becomes erroneously negative or undefined if: S B + h ≤ (7.26) b 2
If this condition arises, the model incrementally adds 0.5 meters to the berm height B until (7.26) is untrue. The beach retreat E∞ associated with this adjusted berm height is computed rather than using the height that you provided. The output report produced by the model will notify you that the berm height has been adjusted and by how much if this is the case. The scale coefficient α (π ≤ α ≤ 2π) is computed by solving the following equation:
exp(−α/β) = cos α˘(1/β) sin α (7.27)
7.3. The model 94 InVEST User Guide, Release 2.5.6
where β is a function of the finite storm duration Td and breaking wave characteristics: 2π H3/2 h mx −1 b b b (7.28) β = 320 √ 3 1 + + Td gA B + D hb
Practically, the model estimates the amount of beach retreat that would occur under various management scenarios by first solving Equation (7.24) in the absence of vegetation. Breaking location is computed as explained above, using the sediment scale factor A derived from the sediment size that you input. In the presence of vegetation, it is often difficult to estimate the exact location of breaking, and there is not any guidance or observation of avoided beach retreat in the presence of natural habitats. Consequently, the amount of beach retreat in the presence of natural habitats is estimated by scaling the amount of retreat obtained in the absence of natural habitats by the ratio of reduction in runup values as well as the ratio of the cube of wave height over the submerged vegetated bed. This is because empirical models of beach retreat are directly proportional to water level (e.g., see Equation (7.25)). Also, process-based models of beach erosion (e.g., Kriebel and Dean, 1985) scale erosion by wave dissipation, which is proportional to the cube of wave height. The model’s final output value of erosion in the presence of natural habitat is the average of both values. Note: You may notice that for certain values of m, Equation (7.25) can yield negative results. Instead of generating a message error, the profile foreshore slope is decreased so that E∞ is positive. This correction is made because of the uncertainty associated with the erosion model and your inputs. In future versions of this model, a more sophisticated erosion model will be used to avoid this situation. To estimate a correct foreshore slope that won’t yield negative values in Equation (7.25), the model approximates the breaking wave height by using Equation (??) (see Profile Generator Model). Then the model computes the breaking position and depth xb and hb by assuming that Hb = 0.78hb and: 2/3 hb = Axb (7.29)
If the model does adjust the profile slope, be cautious of comparing retreat values to values obtained at neighboring locations or at the same site for other forcing conditions. An increase in slope causes an increase in retreat not associated with increased forcing or the lack of protective habitats. In addition to sandy beaches, the model can also estimate the volumetric erosion a consolidated bed might experience. Muddy substrates, such as those found in marshes or mangrove forests, do not erode in the same manner as sandy beaches. They are composed of cohesive sediments that are bound by electro-magnetic forces, and their resistance to wave- and storm-induced bed velocity is a function of their composition and level of consolidation. In the erosion −1 model, the hourly rate of scour of a consolidated bed Em[cm.h ] is estimated by following the method proposed by Whitehouse et al. (2000, Ch. 4): ( 36(τo − τe)me/CM if τo − τe > 0 Em = 0 if τo − τe ≤ 0
where me is an erosion constant and CM is the dry density of the bed. Both constants can be obtained from site-specific −1 measurements. However, the Erosion Protection Excel Table offers sample default values of me = 0.001m.s and −3 CM = 70kg.m . The variable τe is the erosion shear stress constant (the maximum shear stress the consolidated bed can withstand before sediment begins to scour) and is computed as:
E2 τe = E1CM (7.30)
where E1 and E2 are site specific coefficients. The erosion threshold value within the model has be prescribed using −6 average values of those coefficients (Whitehouse et al., 2000): E1 = 5.42 · 10 and E2 = 2.28. Finally, the wave-induced shear stress τo is computed as: 1 τ = ρf U 2 (7.31) o 2 w bed
7.3. The model 95 InVEST User Guide, Release 2.5.6
where Ubed is the wave-induced bottom velocity at water depth h: p Ubed = 0.5H g/h (7.32)
and fw is the wave-induced friction coefficient, computed assuming the flow is turbulent:
σU 2 −0.187 f = 0.0521 bed (7.33) w ν where ν ≈ 1.17 · 10−6m2.s−1 is the kinematic viscosity of seawater, and σ = 2π/T is the wave frequency. The model estimates the rate of bed erosion for regions that are above MLLW, assuming that there is no mixture of sand and mud in the inter- and supra-tidal areas. Since the wave height H and, therefore velocity at the bed Ubed decays from the shoreline moving inland, the model is able to compute the spatial variation of the scour rate with respect to distance from the shoreline. By integrating under the spatially varying scour rate curve and multiplying by the duration of the storm, the model also yields an approximate of the volumetric sediment loss at along the modeled profile. The model also returns the distance inland where erosion is expected based on the inland limit of where the bed shear stress exceeds the threshold value. Further, since the reduction in habitat footprint and/or density will increase wave heights and, therefore, scour rates, the model computes the spatially varying scour rates and volumetric sediment loss for the present and modified habitat footprints. In other words, the model estimates the increase in erosion due to the removal or modication of natural habitats.
Valuation
The Erosion Protection model quantifies the protection provided by habitats in terms of the avoided damages to prop- erty due to erosion from waves. The market value of properties in the area, based on a sample of recent sales, can be used as an estimate of property value. Tax assessment data and replacement cost methods are the two other common ways to quantify the value of properties. Any of these three options are valid inputs to this model. As this model is intended to work in data-poor regions, you only need to provide information about average property value in a given area. However, the extent to which this will reflect the true value of these properties, and thus the value of habitat in providing protection from storms, directly depends on the quality of the property value input. For more information on the merits of each data source and a general discussion of valuation using this approach, please refer to Cannon (1995). Coastal storms damage properties, and the difference in damages due to habitat can be considered an indicator of the value of those habitats in providing protection to properties from storms. The main inputs to the valuation module are the areas of erosion from the baseline and management scenarios, avoided land loss amount, and local data on property values that you provide. The difference in distance eroded between the baseline and management scenarios is referred to as “avoided erosion” (RA) or avoided land loss:
RA = R2 − R1 (7.34)
where R1 and R2 are estimates of the total amount of land area eroded under habitat scenario one (baseline) and habitat scenario two (management scenario) over a longshore distance L over which outputs of the waves and erosion are valid. We leave it up to you to define this longshore distance L. However, in general, it can be defined as an area where bathymetry, topography and natural habitat characteristics do not vary much. The values of land eroded Rx under each habitat scenario, x = 1, 2, is obtained from the erosion outputs of the erosion model Ex as:
Rx = ExL (7.35)
7.3. The model 96 InVEST User Guide, Release 2.5.6
As changes in land use need to be considered against other possible investments and time preferences, it is appropriate to consider the expected present value, EPV , of services provided by habitat. The EPV calculation employs a discount rate, i, over a time horizon that you define, τ, expressed in years. EPV reflects the value of the stream of avoided storm damages over time due to a change in habitat and discounts the value of those avoided damages in distant periods when the discount rate is greater than zero. We have provided a default discount rate, but you should assess whether that is appropriate for your case. For more information, see this website. The EPV is also a function of the expected return period associated with your storm. Storms are classified by their strength and probability of occurrence. Thus, it is common to hear them referred to as a “hundred-year storm” or a “thousand-year storm,” where the expected frequency of a storm of that strength is once per hundred or thousand years. The annual probability of occurrence for a storm that occurs on average every T years is p = 1/T , where p is constant across time - that is, not contingent on previous occurrences. The model estimates the value of habitats for coastal protection from erosion as the difference in damages under two habitat scenarios, given as DA = D2 − D1 for a given storm class with an expected return time of T . Because storms occur at irregular intervals over time, the model allows you to assess these benefits across a defined time horizon. EPV for a given storm class is calculated as:
τ X pDA EPV = (7.36) (1 + i)t t=1
7.4 Limitations and Simplifications
Although the Tier 1 Erosion Protection model will help you inform management decisions by demonstrating the protective capabilities of natural habitats, it has limitations (theoretical and otherwise). A primary limitation is that the Erosion Protection model assumes that all erosion leads to a loss of land. In some places this assumption will reflect reality; in other locations, erosion from large storms can be reversed through net sediment accretion during periods of calm weather. These sorts of more complex physical dynamics are beyond the scope of the model and analysis we present here. Further, the model estimates coastal protection services provided by habitats in terms of the reduction in damages due to erosion from storm waves, not surge. Some coastal habitats have the ability to attenuate surge in addition to waves (e.g., marshes, coastal forests), while other nearshore subtidal habitats do not (e.g., eelgrass). If you are modeling storm waves from hurricanes which also produce significant surge, the current model likely underestimates the protection value of the former habitats. In addition to the limitations discussed the above, the model has technical limitations. The first is the lack of high quality GIS data that are readily available. In the event that you do not have a nearshore profile for the region of interest, simple rules of thumb based on observations are provided to help you generate one. Though grounded in the literature, these rules of thumb will not generate profiles that perfectly match all sites of interest. Again, a site visit to obtain missing data will improve the generated profile, and thus the model results. The theoretical limitations of the Nearshore Waves and Erosion model are more substantial. As mentioned earlier, wave evolution is modeled with a 1D model. This assumes that the bathymetry is longshore-uniform (i.e. the profile in front of the site is similar along the entirety of the stretch of shoreline). Because this is unlikely true, the model ignores any complex wave transformations that occur offshore of the site of interest. Also, although the wave model used compares well against observation with default calibration parameters (see Nearshore Waves and Erosion) you are not currently offered the option to calibrate it. Thus, values of wave height and wave-induced water level along the modeled transect might differ from observations. Another limitation of the wave model is that it has been assumed that the vegetation characteristics that you provide in the Erosion Protection Excel Table remain valid during the storm forcing that is being modeled. The model also ignores any non-linear processes that occur when waves travel over submerged vegetation. For example, the model does not take into account wave reflection that occurs at the edge of the vegetation field, motion of vegetative elements caused by wave forces, or reductions in habitat density that might occur during a storm. Furthermore, default values of friction and drag coefficient are used to compute the forces exerted by the habitats on the water column. This implies
7.4. Limitations and Simplifications 97 InVEST User Guide, Release 2.5.6 that those forces are independent of the flow turbulence regime. Finally, simple empirical models are used to compute the wave profile over coral and oyster reefs. Although these models have been validated with observations, they ignore many processes that might change the wave profile that the model computes. You should also be aware that, while under some small levels of storm surge oyster reefs provide some wave protection, the primary role of oyster reefs is to prevent wave erosion of saltmarshes during typical or day to day wave conditions and water levels. To model beach erosion, the model proposed by Kriebel and Dean (1993) is used. Although this empirical model has been widely used (USACE, 2002), it ignores key erosion processes that occur during a storm. For example, the dynamic response and feedback between waves and the bed profile during the storm is not taken into account. The model also does not evaluate when dune breaching and the amount of overwash that might occur during the simulated storm. To model scour of consolidated beds, the model proposed in Whitehouse et al. (2000) is used, and, in the Erosion Protection Excel Table, default sediment characteristics are provided but are not appropriate for all sites. Further, the assumption that the whole bed has the same characteristics, both horizontally and vertically, is made. Finally, any dynamic response between increase levels of suspended sediments and wave-induced bottom velocity, as well as any sediment redeposition and settlement, are ignored. Site-specific input parameters might help improve the accuracy of model results relative to using the provided default parameters, but will not compensate for the physical simplifications made. The avoided damages model can be categorized as a partial equilibrium analysis. It investigates the erosion protection value of habitat with a process-based model by changing only the level of habitat, holding all else constant. However, it is important to realize that a dramatic change in habitat management may decrease the reliability of results. For example, you could make the case that property values would change in response to a large shift in habitat, invalidating our assumption that these values stay constant in the baseline and proposed scenario. The expected reduction in erosion due to habitats calculated for the model is for a storm of the size modeled in the wave evolution and nearshore erosion modules. However, during the time period defined by you, the coastal habitats will likely provide protection against a wide range of different sized waves and storms. The most accurate way to value habitats for their total protection services would be conduct multiple runs of the model for different sized storms that occur with different return periods (e.g., different hurricane categories) and add together the expected avoided damages due to habitat protection outside of the model, such that:
τ1 τ2 τ2 X pStorm1DA1 X pStorm2DA2 X pStorm2DA2 EPV = + + + ... (7.37) AllStorms (1 + i)t (1 + i)t (1 + i)t t=1 t=1 t=1
You can perform this calculation by adding together the results from multiple runs of the InVEST model for different sized storms with different return periods. Finally, this model does not explicitly account for property owners’ behavioral response to erosion over time. De- pending on this response, the assessed value may overstate or understate the potential damages. As an example of this behavioral response, communities may erect manmade erosion protection, property owners may raise their houses, many may sell their property and hence property values will decrease, etc. In summary, the interactions between waves and the shoreline represent extremely complex processes. Calculating avoided damages from erosion using data that aggregates property value (e.g., land value and structure value) into one value simplifies the diverse and complex ways individual properties are damaged by erosion (e.g., loss of land versus damages to structures). Thus, our estimates of the value of coastal protection services do not distinguish between these possibilities. The simple model presented here is designed to capture the essence of these and to guide your understanding of the roles that nearshore habitats might play in mitigating the coastal hazards of erosion and inundation.
7.4. Limitations and Simplifications 98 InVEST User Guide, Release 2.5.6
7.5 Data Needs
As mentioned earlier, the Erosion Protection model is composed of two primary sub-models: the Profile Generator and the Nearshore Waves and Erosion models. It is recommended that you first utilize the Profile Generator tool to obtain a cross-shore profile that contains bathymetry and backshore information. This tool will also help you obtain several pieces of useful information including: the bathymetry and nearshore topography along the profile of interest; the type of natural habitats present at the site, as well as your location along the profile; values for offshore wave height and wind speed and fetch direction for the site. Once this profile information has been obtained and forcing parameters have been selected, you can run the Nearshore Waves and Erosion model. Also, to investigate the impacts of management actions on waves and erosion, you can select the type of management action or change the footprint and density of each habitat. Running the Nearshore Waves and Erosion model requires, at a minimum, a bathymetry profile as well as wave and storm information. Furthermore, information on the type of backshore present at the site, as well as on the characteristics of the natural habitats that are present at the site will be needed.
7.5.1 Profile Generator
1. Workspace (required). You need to specify a workspace folder path where the model outputs can be stored. It is recommended that you create a new folder that will contain all CP Tier 1 outputs (Profile Generator as well as Nearshore Waves and Erosion outputs). For example, by creating a folder called “WCVI” inside the “Coastal- Protection” folder, the model will create “_Profile_Generator_Outputs” and/or a “_NearshoreWaveErosion” folders containing outputs from your various runs, as well as an intermediate folder named “scratch”. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\CoastalProtection\WCVI
2. Label for Profile Generator Run (10 characters max) (required). Provide a short name that reflects the loca- tion or reason of your run. This name will be used to create a subfolder inside the “_Profile_Generator_Outputs” folder that will contain outputs for your model runs. For example, if you chose the label “Dune_2m” because you wanted to see what a cross-shore profile with a 2m dune looked like, a folder called “Dune_2m” inside the “_Profile_Generator_Outputs” folder will be created. That folder will contain two subfolders called “html_txt” and “maps”. The “html_txt” folder contains an html file that summarizes information about the site of interest with figures of the created profile and showing the location of natural habitats along the profile. The “maps” folder contains shapefiles that can be viewed in GIS. These shapefiles include polylines that show fetch vectors and fetch distances, points along the transect where topo/bathy was extracted as well as points showing the locations of natural habitats. Name: A concise label describing the model run File type: text string (direct input to the ArcGIS interface) Sample: Dune_2m
3. Land Point (required).. Provide a point shapefile of the location where you want to run the Profile Generator. The datum of this input must be WGS 1984 in order to avoid transformations when the model projects the Wave Watch III Model Data input. It is highly recommend that you use snapping to ensure that the point is on the edge of the land polygon (shoreline). From this location the Profile Generator will extract a profile orthogonal to the land (if you are cutting a transect in GIS), gather wind and wave data from the closest deep-water WW3 grid point, and/or compute fetch distances, averaged over 16 directions. If you are cutting a cross-shore transect in GIS, make sure to inspect the coastline around this input and adjust the Land Point Buffer Distance (input 8) accordingly. Name: File can be named anything, but no spaces in the name File type: point shapefile (.shp) Sample path (default): \InVEST\CoastalProtection\Input\LandPoint_BarkSound.shp
4. Land Polygon (required). This input provides the model with a geographic shape of the coastal area of interest, and instructs it as to the boundaries of the land and seascape.
7.5. Data Needs 99 InVEST User Guide, Release 2.5.6
Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path (default): \InVEST\Base_Data\Marine\Land\LandPolygon_WCVI.shp
5. Do you want us to cut a cross-shore transect in GIS? (required). This drop down box allows you to select whether you 1) wish to have the GIS create a cross-shore transect, 2) will upload a cross-shore profile of your own or 3) prefer to have the model create a theoretical profile. The answer provided to this question will determine whether subsequent inputs are required or optional. File type: drop down options Sample: (1) Yes
6. Bathymetric Grid (DEM) (optional). If you have answered “(1) Yes” to the question: “Do you want us to cut a cross-shore transect in GIS?”, the model requires a DEM in order to cut a cross-shore profile. This bathymetric grid layer should have a vertical elevation referenced to Mean Lower Low water. Name: File can be named anything, but no spaces in the name File type: raster dataset Sample path: \InVEST\Base_Data\Marine\DEMs\claybark_dem
7. Habitat Data Directory (optional). If you have answered “(1) Yes” to the question: “Do you want us to cut a cross-shore transect in GIS?”, the model will optionally allow for the location of natural habitats that intersect on the cross-shore transect. To do so, you must store all Natural Habitat input layers that you want to consider in a unique directory. Each natural habitat layer should consist of the location of those habitats, and all data in this folder must be polygon shapefiles and projected in meters. Further, each of these layers should end with an underscore followed by a unique number, for example “_1” or “_2”. The model allows for a maximum of six layers in this directory. Do not store any additional files that are not part of the analysis in this folder directory. If you need to add or remove natural habitat layers at one site for various analyses, you will have to create one “Natural Habitat” folder per analysis (omitting the habitat you wish to remove). If you wish to exclude natural habitat from your analysis, simply leave this input blank. Name: Folder can be named anything, but no spaces in the name File type: None, but must contain polygon shapefiles (.shp) Sample path: \InVEST\CoastalProtection\Input\NaturalHabitat
8. Land Point Buffer Distance. If you have answered “(1) Yes” to the question: “Do you want us to cut a cross- shore transect in GIS?”, the model requires this distance value in order to create a perpendicular transect based upon the slope of the coastline near the Land Point (input 3). The Land Point shapefile must be within this buffer distance from the shoreline as defined by the Land Polygon (input 4). Also, the terrestrial area located behind or in front of that point must be wider than the buffer distance. In general, a distance of 250m is sufficient. However, if the site is along a narrow island or a spit that distance should be smaller than the width of the island or the spit. It is recommended that if your Land Point is placed near a sinuous coastline (e.g. surrounded by narrow inlets), you should determine the maximum distance from the Land Point in both directions along the coast without crossing an abrupt change in angle of the coastline. This distance measure should be entered as the Land Point Buffer Distance and will allow the model to determine the true angle for a transect perpendicular to this Land Point site. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 250
9. Length of your profile. If you have answered “(1) Yes” to the question: “Do you want us to cut a cross-shore transect in GIS?”, the model requires the length of the profile you wish to create from the Land Point (input 3) to a suitable offshore limit (in km). If the provided DEM is seamless, the Profile Generator extracts topography for the same length inland of the point. This length should be the distance from the Land Point to the deepest adjacent point (in a sheltered region or in an estuary) such that an adjacent land masses are not intersected, or to a sufficiently deep point along an open or exposed coastline.:
7.5. Data Needs 100 InVEST User Guide, Release 2.5.6
Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 25
10. Cross-Shore Profile (optional). If you have answered “(2) No, but I will upload a cross-shore profile” to the question: “Do you want us to cut a cross-shore transect in GIS?”, the model will not cut a cross-shore profile for you from a GIS layer, but will create a smooth backshore profile, or manipulate a cross-shore profile of your choice. This file must contain a minimum of 2 (X,Z) coordinates. It must be tab delimited with two columns. The first column must be the cross-shore distance X-axis, where X=0 is at the shoreline (positive X pointing seaward, negative X pointing landward). The spatial resolution of the X-axis (spacing between two X-coordinates) must be equal to 1 (dx=1). The second column must indicate the cross-shore elevations along the X-axis. Depths values must be negative (referenced to Mean Lower Low Water) and terrestrial elevations must be positive.: Name: File can be named anything, but no spaces in the name File type: Tab delimited text file with two columns (X,Z) (.txt) Sample path: \InVEST\CoastalProtection\Input\Depths.txt
11. Smoothing Percentage (required). Enter a percentage value for how much you wish to smooth the profile created or fed through the model. A value of “0” means no smoothing. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 5
12. Erosion Protection Excel Table (required). This file contains information about your site that will allow the model to build a full cross-shore profile, including tidal elevations, and profile slope modifications. Also, the locations of natural habitats will be populated here by the Profile Generator Model if you include the Habitat Data Directory as input. This table has 4 sections: General Site Information, Foreshore/Backshore Profile Modifications, Habitats, and Habitat Management Action. Three of the sections, General Site Information, Foreshore/Backshore Profile Modifications, and Habitats are applicable to the Profile Generator tool. In the Foreshore/Backshore Profile Modifications section, you have the option of modifying the topo/bathy profile by inserting linear slopes along the profile. You are required to populate the Habitats section if you include a Habitat Directory in the Profile Generator model. For more information on how to complete this Erosion Protection Excel Table, please see Erosion Protection Excel Table. Name: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if you have MS Excel 2007 or newer) Sample path: \InVEST\CoastalProtection\Input\ErosionProtection_WCVI_BarkSound.xls
13. Wave Watch III Model Data (optional). If you would like the model to gather wind and wave statistics that might represent oceanic conditions at your site, upload the WW3 file that has been provide in the InVEST download package. The model will use this dataset to read the maximum, top 10% and top 25% wind speed as well as wave height and associated wave period values from the model grid closest to your site. Name: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\CoastalProtection\Input\WaveWatchIII.shp
14. Wave Watch III Search Distance (kilometers). The model requires this search distance in order to find the closest WW3 point. The default distance is 50 km, but may need to be increased depending on the distance of your Land Point to the nearest WW3 point. To determine the appropriate distance for your site, use ArcGIS to measure the distance (over water) of the Land Point to the nearest WW3 Model Data point. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 50
7.5. Data Needs 101 InVEST User Guide, Release 2.5.6
15. Do you wish to calculate fetch for Land Point? (optional). This drop down box allows you to specify whether you want the model to compute fetch distances. If “(1) Yes” is selected, fetch radials will be extended from the Land Point (input 3) and cut based on the Land Polygon (input 4). The results will be averaged over 16 directions. File type: drop down options Sample: (1) Yes
7.5.2 Nearshore Waves and Erosion
The Nearshore Waves and Erosion model estimates the profile of wave height over your bathymetry from an offshore value to the shoreline. It is used to estimate the amount of erosion of a beach or a muddy substrate. This section explains how to obtain and/or interpret all the data the model requires to run properly. 1. Workspace (required). You need to specify a workspace folder path where model outputs will be stored. It is recommend that you input the same workspace folder that you input in the Profile Generator, which will contain all CP Tier 1 outputs (Profile Generator as well as Nearshore Waves and Erosion outputs, see Profile Generator). In this workspace, we will create a folder name “_WaveModel_Outputs” that will contain all Nearshore Waves and Erosion outputs. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\CoastalProtection\WCVI
2. Label for Waves and Erosion Run (10 characters max) (required). Provide a short name that reflects the reason for your run. This label will be used as a suffix to all outputs created inside the “_WaveModel_Outputs” folder. For example, if you chose the label “Dune_2m” to evaluate the protective services provided by a 2m sand dune, the model will create an html output file named “OutputWaveModel_Dune2m” as well as a text file indicating wave height as a function of cross-shore distance named “WaveHeight_Dune2m” Name: A concise label describing the model run File type: text string (direct input to the ArcGIS interface) Sample: Dune_2m
3. Erosion Protection Excel Table (required). You are required to fill out and upload the Erosion Protection Excel Table. This table contains information about tide levels, the type of substrate at your site, the type and physical characteristics of natural habitats, and how the management action affects the natural habitats. For more information on how to complete this Erosion Protection Excel Table, please see Erosion Protection Excel Table. Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if you have MS Excel 2007 or newer) Sample: InVEST\CoastalProtection\Input\ErosionProtection_WCVI_BarkSound.xls
4. Cross-Shore Profile (required). A cross-shore profile is required (which can be obtained from the Profile Generator’s outputs) in order to model wave height evolution in your area. The output text file can be found in the “html_txt” folder of a successful PG run and will be called “CreatedProfile_[suffix].txt”. This file must contain a minimum of 2 (X, Z) coordinates, and must be tab delimited with two columns. The first column must be the cross-shore distance X-axis, with X=0 at the shoreline (positive X pointing seaward, negative X pointing landward). The spatial resolution of the X-axis (spacing between two X-coordinates) must be equal to 1 (dx=1). The second column must indicate the cross-shore elevations along the X-axis. Depth values must be negative (referenced to Mean Lower Low Water) and terrestrial elevations must be positive. Name: File can be named anything, but no spaces in the name File type: Tab delimited text file with two columns (X,Z) (.txt) Sample path: InVEST\CoastalProtection\WCVI\_ProfileGenerator_Outputs\Dune_2m\html_txt\CreatedProfile_Dune_2m.txt
7.5. Data Needs 102 InVEST User Guide, Release 2.5.6
5. Do you have wave height and wave period values? (required) The model requires the wave height and period at the offshore edge of your profile as starting conditions. This drop down box allows you to select whether you 1) will provide wave height and wave period values or 2) will instead provide wind speed, fetch distance, and water depth. If you choose answer 1: “Yes, I have these values”, enter them below the prompts starting with “IF 1:”. If you choose answer 2: “No, please compute these values from wind speed and fetch distance”, enter a wind speed, fetch distance as well as average water depth at your site below the prompts starting with “IF 2:”. If you have run the Profile Generator and input WW3 data and had the model compute fetch distances for you, you can use that model run’s html outputs for default values of wave height and period, wind speed and fetch distances. Figures 12 and 13 can also be used as guidance for typical wave height and wind speed observed during certain classes of storms. File type: drop down options Sample: (1) Yes
6. Wave Height (meters) (optional).: Wave height is the distance between the wave crest and wave trough, as shown in the figure under Fetch Distance (below). For typical values of wave period during storms, see the following figure. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface)
Figure 7.2: Typical values of wave height and associated wave period for various types and classes of storms. Use this information to make the best possible guess of wave characteristics offshore of your site.
7. Wave Period (seconds) (optional).: Wave period is the amount of time, in seconds, necessary for two consec- utive wave crest to pass a fixed point (see the figure under Fetch Distance below). Wave period should be less than 20s. For typical values of wave period during storms, see the preceding figure. Name: A numeric text string smaller than 20 seconds (positive integer) File type: text string (direct input to the ArcGIS interface)
8. Wind Speed (meters per second) (optional).: Strong winds blowing steadily over the water can generate high waves if the fetch distance is long enough. Please enter a wind speed value that is representative of the conditions that you want to represent at your site. Please remember that wind patterns at your site might have a seasonal
7.5. Data Needs 103 InVEST User Guide, Release 2.5.6
signature and vary depending on the direction they blow towards. If you have uploaded WW3 data in the Profile Generator, we provide you in the html output a wind rose representing typical storm wind speeds at your site, coming from 16 equiangular directions. Also, the following figure can also be used as a guidance for typical wind speed observed during certain classes of storms.: Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface)
Figure 7.3: Typical values of central pressure, wind speed and surge level for various classes of hurricanes. Use this information to make the best possible guess of wind speed offshore of your site if you want the model to estimate values of wind-generated wave height and period during your storm. Also, use this information to make the best possible guess of surge elevation during your storm.
9. Fetch Distance (meters) (optional).: Fetch is defined here as the distance travelled by winds over water with no obstructions, for a certain compass direction. Winds blowing over a longer fetch generate higher waves than winds blowing over a smaller fetch distance. You can get fetch directions for the 16 equiangular directions that form a compass by choosing the fetch option in the Profile Generator tool (see the following figure). Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface)
Figure 7.4: Definition of various coastal engineering terms used in the model.
10. Water Depth (meters) (optional).: For a given fetch distance, wind blowing over a shallow area generate smaller waves than wind blowing over the deep ocean. Here, enter the average depth value along the fetch angle
7.5. Data Needs 104 InVEST User Guide, Release 2.5.6
that you have chosen (see the preceding figure). This value will be used to generate realistic values of wave height and associated period at your site. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface)
11. Storm Duration (hours) (required).: In order to estimate the amount of beach erosion or bed scour in inter- and/or supra-tidal areas, enter the maximum water level reached during your input storm, as well as its duration. Please indicate the duration of the storm you wish to model. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 5
12. Surge Elevation (meters) (required).: In order to estimate the amount of beach erosion or bed scour in inter- and/or supra-tidal areas, enter the maximum water level reached during your input storm. Please make sure that the storm surge level you input is consistent with the wind speed or wave height that you entered. For guidance, please consult the Wind Speed figure for storm surge levels typically observed during hurricanes. This surge elevation is applied to the MSL. If you want to investigate, for example, a storm hitting your area at high tide you must add the high tide elevation to this surge value and enter the sum for this input. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 1
13. Model Spatial Resolution (dx) (required): A coarse spatial resolution can sometimes lead to model instability and inaccuracy in model outptuts. Please choose a proper resolution at which you want to run the model. This value can be greater or smaller than one. However, keep in mind that a smaller resolution results in longer computing time. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 1
14. Compute Econonomic Valuation (optional): By checking this box, users will instruct the model that they would like to approximate a monetary value for habitat and the loss in this value owing to habitat modification (reduction). 15. Longshore Extent (meters) (required for valuation): To obtain an approximate area of land loss associated with retreat/erosion, the retreat/erosion distance must be multiplied by a longshore length. Essentially, this is the length along the shore where one would expect the same amount of retreat. In other words, this is the along shore length where the natural habitat types, coverage, and management actions, as well as, topo/bathy and forcing conditions are approximately uniform. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 250
16. Property Value (local currency) (required for valuation): This is the average monetary value of the land, per square meter, that you wish to use in the valuation computation. This model is envisioned as a way to calculate property damages due to erosion, however you could use any source of per-area value. For example, if you had data on the replacement cost of beaches per square meter this model could provide estimates of the avoided replacement costs due to habitat. Value enters into the model as a function of area eroded, so it can be used to measure any valid loss in value due to erosion.: Name: A numeric text string (required for valuation) File type: text string (direct input to the ArcGIS interface)
7.5. Data Needs 105 InVEST User Guide, Release 2.5.6
17. Return Period of Storm (years) (required for valuation): This is the number of years between occurences of the storm (surge and waves) applied in the model run that is experienced at your site. More extreme storms are more infrequent than less extreme storms. Typical return period used in risk assessment are 10, 50, 100, and 500 years, with 10 years being the most common and mild conditions and 500 years being very extreme and infrequent/less likely storm conditions. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 10
18. Discount Rate (required for valuation): A discount rate to adjust the monetary benefits of the natural habitats in future years to the present time is required. We provide a default value of 5%, however you are strongly encouraged to evaluate the appropriate rate for your decision-making context.: Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 0.05
19. Time Horizon of Valuation (years) (required for valuation): This is the number of years over which you intend to value the coastal protection services provided by your habitat. Name: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface)
Erosion Protection Excel Table
The Erosion Protection Excel Table contains four sections: General Site Information; Profile Modification; Habitats; and Habitat Management Action. General Site Information 1. Tidal Elevations: You are to enter the elevation of Mean Sea Level (MSL) and Mean High Water (MHW) relative to Mean Lower Low Water (MLLW). Since most bathymetric/nearshore surveys are conducted at the lowest tides, it has been assumed that the vertical datum of the source of bathymetry data (DEM, text file corresponding to an actual cross-section survey, etc.) is MLLW. If it is known that the vertical datum of the bathymetry data is something other than MLLW, enter the elevation of MSL and MHW relative to the known datum. For example, if the vertical datum is actually MSL and the elevation of MHW above MSL is 0.5 m, a value of 0 and 0.5 should be entered in MSL and MHW columns, respectively. In the example shown in the screenshot below, the topo/bathy elevations are presumed to be relative to MLLW, and MSL and MHW are 0.3 m and 0.6 m above MLLW, respectively.
Figure 7.5: Screenshot of the Tide Information fields within the “General Site Information” section of the Erosion Protection Excel Table.
This information is used by the Wave and Erosion Model to shift the profile depths to be relative to MSL. Also, a link is provided in the Excel table to a figure showing tidal ranges (the difference between MHHW and MLLW elevations) from around the world. If you are uncertain of the tidal elevation values you have entered, you can check this figure to ensure if the tidal range agrees with the values that you have entered. Otherwise, you can approximate MSL as half the value of the tidal range and MHW as the value of the tidal range shown in this figure. As with all inputs, if accurate local measurements of tides are available, these data should be used.
7.5. Data Needs 106 InVEST User Guide, Release 2.5.6
Figure 7.6: Worldwide variation of tidal range. This information can be used to make the best possible guess of tide elevation at the site of interest.
2. Type of backshore, Sediment and Beach Characteristics: Here, you define what type of sediments make up your backshore. Please refer back to cp-ProfOptions for a more complete description of the two options. Option number 1 corresponds to a sandy backshore and option 2 corresponds to a muddy backshore; this tells the Wave and Erosion Model which erosion computation to run. You must also enter the median diameter or size of the sediment at your site. If you have a qualitative description of the sediment at your site (coarse sand, very fine sand, silty, etc.), a representative sediment size can be obtained by using the Unified Soil Classification (from Dean and Dalrymple, 2002, Ch. 2) shown below; a link to this figure is contained in the Excel table. If the sediment size does not correspond to the backshore option (if Option 1, sandy beach is selected and the sediment size corresponds to clay/mud, for example), an error message lets you know that you must change the sediment size to agree with the backshore option. If the option is a sandy beach and a valid sediment size is entered, you are required to enter the following characteristics of your sandy beach: dune height, berm width, berm elevation, and foreshore run. The dune height is the elevation of the dune crest relative to the berm. If you are unsure whether or not dunes exist at your site, a map showing the worldwide distribution of dunes is provided and is shown below. The berm width is the width of the sandy beach from the shoreline to the toe of the dune or other backshore feature (coastal development, estuary, etc.). The berm elevation is the elevation of the sandy beach relative to MSL. It is recommended that the berm elevation be at least as high as the elevation of MHW. Lastly, the foreshore run is the inverse of the foreshore slope. The Excel table populates suggested foreshore runs for you to choose from based on sediment size. The figure below shows a pictorial definition of these characteristics of a sandy backshore. Since berm height and width, as well as dune elevation, is easily obtained from visual estimates, we encourage you to visit your site to obtain the most accurate values for these parameters. Also, the foreshore slope can be obtained from a simple survey method, see Appendix A. If the option is a muddy system and a valid sediment size is entered, the model requires a dry density value and an erosion constant for the sediment at the site. These parameters cannot be approximated by visual observations or simple methods but require laboratory testing of site samples. Therefore, default values are provided in the Excel table. If you have these values specifically for your site or region, you can overwrite these defaults. In addition to informing the Wave and Erosion Model about which erosion models to run, and the important physical characteristics for those models, this information also informs the Profile Generator. For example, if a sandy beach is selected, the Profile Generator will incorporate the beach geometry (foreshore slope, berm height and width, and dune
7.5. Data Needs 107 InVEST User Guide, Release 2.5.6
Figure 7.7: Sediment size classification. Use the table and Geotechnical Gauge to make the best possible guess of sediment size at the site.
Figure 7.8: Map showing the approximate distribution of sand dunes in the world. This information can be used to make a guess about whether or not there’s a sand dune at the site.
7.5. Data Needs 108 InVEST User Guide, Release 2.5.6
Figure 7.9: Definition of Berm Height, Berm Width and Dune Height at a typical sandy beach.
7.5. Data Needs 109 InVEST User Guide, Release 2.5.6 height) into the generated profile. Also, if you opt for the Profile Generator to create an Equilibrium Beach Profile (for sandy beaches only), the Profile Generator uses the sediment size provided here to compute the sediment scale factor (see Equation EqProf ). The figure below is a screen capture of where you enter these backshore and sediment characteristics. In the example below, the site is a sandy beach with medium sized sand. Since the option and sediment corresponds to a sandy beach, the dry density and erosion constant fields are greyed out. If this example corresponded to a muddy system, the sandy beach fields would be greyed out and the dry density and erosion constant fields would appear.
Figure 7.10: Screenshot of the backshore and sediment characteristic fields within the “General Site Information” section of the Erosion Protection Excel Table.
Profile Modification In this section, you can superimpose three linear (monotonic) segments onto their topo/bathy profile. To add a mono- tonic profile, the run value “R” (slope=1/R) as well as the cross-shore locations between which this monotonic slope will apply are required. For a flat profile, you can either enter 0 or a very large number. The convention used is that the beginning point of the transect is seaward of the end point. Also, the origin of the X-axis is at the shoreline, with positive X pointing offshore, and negative X pointing landward of the shoreline. In the screenshot shown below, the user wishes to place a slope of 1/600 from the shoreline to 5 km (5000 m) onshore.
Figure 7.11: Screenshot of the “Profile Modification” section of the Erosion Protection Excel Table.
This example likely corresponds to a case where the elevation was not seamless, or there was no topography measure- ment, and the user is applying a typical slope associated with mangroves as the backshore profile. Habitats In this table, you indicate the types of natural habitats that are present in the Natural Habitats folder that was specified in the Profile Generator prompt. If you intend to have the Profile Generator place habitats on the cross-shore profile rather than record the locations manually, this table must be filled out. To let the Profile Generator know which layer in the folder corresponds to which habitat type, you will need to enter in the Habitat ID cell the number that corresponds
7.5. Data Needs 110 InVEST User Guide, Release 2.5.6 to the suffix in the shapefile name corresponding to that habitat (e.g., “1”, or “5”, etc.). If a particular habitat is not present, those cells should be blank. In the example below, mangroves, seagrass beds, and coral reefs are present in the study region and the suffix corresponding to these habitats are 1, 2, and 3, respectively
Figure 7.12: Screenshot of the “Habitat” section of the Erosion Protection Excel Table.
Habitat Management Action 1. Sand Dune: If your management action includes reducing the height of your sand dune (or if you would like to investigate the increase in erosion if your sand dune was lowered or removed), you should enter the percent height reduction in this field. A value of 0 corresponds to no change while 100 corresponds to full removal. In the example shown below, the management action is to reduce the height of the dune by 75%.
Figure 7.13: This is where you can define the percent reduction in your dune height associated with a management action.
2. Vegetation: You can specify the physical characteristics of three types of nearshore vegetation: mangroves, seagrass and marshes. You can treat coastal forests as mangroves. For each vegetation type, you need to indicate a representative height, stem diameter and stem density. See the following figure for a definition of those terms, and see the next figure for sample values of these characteristics for seagrass, marshes, and mangroves.
Figure 7.14: Definition of vegetation characteristic terms used in the model.
You also need to indicate the distance of their landward and seaward edges from the shoreline (X=0). In our convention, positive X point offshore, and negative X point landward. All vegetation in inter- and supra-tidal regions should have negative X positions and if positive x-locations are assigned for mangroves or marshes, the model assumes that you intended those values to be negative. If you properly included natural habitat in a Profile Generator run, the
7.5. Data Needs 111 InVEST User Guide, Release 2.5.6
Figure 7.15: Typical example of vegetation characteristics values for the various habitats used in the model.
7.5. Data Needs 112 InVEST User Guide, Release 2.5.6
Pre-Management Action positions will be populated for you but you should double check these values; the Profile Generator may place marsh or mangrove habitats slightly offshore because of differences in projections, precisions, and accuracy of the input layers. Finally, you will have to indicate how they are affected by your management action: • You can change the footprint or location of the vegetation. If the vegetation is completely removed, you should have 0’s for the X locations post-management action. If the footprint is unaffected, the pre- and post- management action footprints should match. • You can also change the density of each vegetation type independently. The model will reduce the density of the habitat for the post-management action by the percentage provided. The following is a screenshot showing the section of the Excel table where the physical characteristics, pre- and post- management locations, and percent density reduction for vegetative habitats are populated. In the example shown, marshes are present from the shoreline (X=0) to 600 meters inland. The marsh footprint is unaffected by the manage- ment action but the density is reduced by 20%. There is also a seagrass bed present from 50 to 500m offshore. The post-management location is reduced to between 50m and 400m offshore but the density is unchanged.
Figure 7.16: A screenshot of the habitat management action section of the Excel table for the vegetation type habitats.
3. Coral Reef: If you have a coral reef at your site, we will evaluate the wave height at its shoreward edge based on its dimensions. First, you need to specify its location along the profile that you uploaded as well as the type of the reef that is present: • If the reef type is a barrier, enter “-1” for both the offshore and shoreward edge locations and “Barrier” for the reef type. • If the reef is located at the shoreward edge of your profile, such as in the case of a fringing reef without a lagoon, the reef location should have the closest distance to shore as 0. The reef type should be defined as “Fringe”. • If the reef is located somewhere along your profile, with a lagoon on its shoreward edge, please enter its location as accurately as possible. The reef type should be defined as “Fringe Lagoon”. Second, you need to specify the physical characteristics of the reef, as defined in the following figure: reef face slope, reef rim slope, depth at reef edge, depth on reef top and width of reef top. Most of these data are obtained through site-specific surveys. However, in case you do not have those data, you can still use our model by entering “0” for the reef face slope, the reef rim slope and the depth at reef edge. You can measure reef width from aerial pictures of your site or from global databases of coral reef (see the Tier 0 Coastal Vulnerability model). Finally, you can enter a best guess for reef top depth knowing that reef top depth values vary between 1 and 2 meters, on average. In this case, we will estimate the wave height on the reef top by assuming that waves break on the reef face, and take an average value for the coefficient Kp in Equation (7.11). Finally, you need to specify how coral reefs are affected by your management action: • If coral reefs are dead but their skeleton is still in place, enter “Dead”. In that case, we will reduce the bottom friction coefficient experienced by waves by half (see Nearshore Waves and Erosion). • If coral reefs are dead and their skeleton failed, enter “Gone”. In this case, we will assume that the reef is now a sandy bottom and adjust the bottom friction coefficient accordingly. • If the reef is not affected by your management action, enter “None”.
7.5. Data Needs 113 InVEST User Guide, Release 2.5.6
Figure 7.17: Profiles of coral reefs in the presence or absence of a lagoon, along with definition of the terms used in the Excel input sheet.
In the screenshot shown below, there is a Fringe Lagoon reef type located from 200m to 500m offshore that will be included in the model. The slopes are unknown but the edge depth, top depth, and top width are 10m, 2m, and 230m, respectively. The management action assigned is “Gone”.
Figure 7.18: An example of inputs for management actions on a Fringe Lagoon reef.
4. Oyster Reef: If you have oyster reefs at your site, you need to enter its distance from the shoreline, as well as its dimensions (see the following figure). If you have a Reef Ball (TM), enter “0” for the crest width. :
Figure 7.19: Depiction of typical shapes of oyster reefs, along with definition of terms used in the input Excel sheet.
7.6 Running the model
7.6.1 Setting up workspace and input folders
These folders will hold all input and output data for the model. As with all folders for ArcGIS, these folder names must not contain any spaces or symbols. See the sample data for an example.
Note: The word ‘path’ means to navigate or drill down into a folder structure using the Open Folder dialog window that is used to select GIS layers or Excel worksheets for model input data or parameters.
7.6. Running the model 114 InVEST User Guide, Release 2.5.6
Exploring a project workspace and input data folder
The /InVEST/CoastalProtection folder holds the main working folder for the model and all other associated folders. Within the CoastalProtection folder there will be a subfolder named ‘Input’. This folder holds most of the GIS and tabular data needed to setup and run the model.
7.6.2 Creating a run of the model
The following example of setting up the Erosion Protection model uses the sample data provided with the InVEST download. The instructions and screenshots refer to the sample data and folder structure supplied within the InVEST installation package. It is expected that you will have location-specific data to use in place of the sample data. These instructions provide only a guideline on how to specify to ArcGIS the various types of data needed and does not represent any site-specific model parameters. See the Data Needs section for a more complete description of the data specified below. 1. Click the plus symbol next to the InVEST toolbox.
2. Expand the Marine, Coastal Protection, and Tier 1 toolsets. There are two scripts that you may want to run in succession: Profile Generator and Nearshore Waves and Erosion. Click on the Profile Generator script to open that model. 3. Specify the Workspace. Click on the Open Folder button and path to the InVEST/CoastalProtection/WCVI folder. If you created your own workspace folder, then select it here. Click on the WCVI folder and click on set the main model workspace. This is the folder in which you will find the “scratch” (intermediate) and “_ProfileGenerator_Outputs” (final outputs) folders after the model is run. 4. Specify the Label for Profile Generator Run. This string of text will be stripped of spaces and shortened to 10 characters. It will serve as the suffix to many of outputs. Type “Dune_2m” into the window. 5. Specify the Land Point. The model requires a land point shapefile to define the location for the analysis.
7.6. Running the model 115 InVEST User Guide, Release 2.5.6
7.6. Running the model 116 InVEST User Guide, Release 2.5.6
Open the InVEST/CoastalProtection/Input data folder. Select the LandPoint_BarkSound.shp shapefile and click to make the selection. 6. Specify the Land Polygon. The model requires a land polygon shapefile to define the land and seascape for the analysis. Open the InVEST/CoastalProtection/Input data folder. Select the LandPolygon_WCVI.shp shapefile and click to make the selection. 7. Select ‘(1) Yes’ that you wish to cut a cross-shore transect in GIS. 8. Specify the Bathymetric Digital Elevation Model (DEM) raster. The model requires a DEM raster file in order to cut a cross-shore transect in GIS. Click and path to the InVEST/Base_Data/Marine/DEMs data folder. Select the claybark_dem raster and click to make the selection. 9. Specify the Habitat Data Directory (optional). The model can use optional polygon shapefile that represents the location of various habitats. Click and path to the InVEST/CoastalProtection/Input data folder. Select the NaturalHabitat folder and click to make the selection. 10. Specify the Land Point Buffer Distance. The model requires this distance order to cut a perpendicular transect in GIS. The default distance is 250 meters, but may need to be modified depending on the site. You may change this value by entering a new value directly into the text box. 11. Specify the Length of your Profile (km). Provide the distance from your land point to a sufficiently deep adjacent location. If the location is sheltered by adjacent land masses, this length should be the distance, from the land point and orthogonal to the land polygon at that location, to the deepest point before crossing any land masses. This parameter defaults to 25 km but is site specific. You may change this value by entering a new value directly into the text box. For this example, 6 km is an appropriate length. 12. Specify the Smoothing Percentage. The model requires this value in order to smooth the bathymetry profile. The default percentage is 5, but may need to be modified depending on the DEM. You may change this value by entering a new value directly into the text box. 13. Specify the Erosion Protection Excel Table. The model requires you to specify information about your site for sediment size, tide elevation and habitats. A sample Erosion Protection Excel Table will be supplied for you. Click and path to the InVEST/CoastalProtection/Input data folder. Double left-click on the file ErosionPro- tection_WCVI_BarkSound.xls. Click to make the selection. 14. Specify the WaveWatchIII Model Data shapefile (optional). The model can use optional wind and wave statistics to represent oceanic conditions at a particular site. Click and path to the InVEST/CoastalProtection/Input data folder. Select the WaveWatchIII.shp shapefile and click to make the selection. 15. Specify the WaveWatchIII Search Distance. The model requires this search distance in order to find the closest WW3 point. The default distance is 50 km, but may need to be modified depending on the distance of your Land Point to the nearest WW3 point. You may change this value by entering a new value directly into the text box. 16. Select ‘(1) Yes’ that you wish to calculate fetch for Land Point. 17. At this point the Profile Generator model dialog box is complete and ready to run. Click to start the model run. The Profile Generator will begin to run and a show a progress window with progress information about each step in the analysis. Once the model finishes, the progress window will show all the completed steps and the amount of time that has elapsed during the model run. 18. Now that your cross-shore profile has been created, you can click on the Nearshore Waves and Erosion script to open that model.
7.6. Running the model 117 InVEST User Guide, Release 2.5.6
7.6. Running the model 118 InVEST User Guide, Release 2.5.6
19. Specify the Workspace. Click on the Open Folder button and path to the InVEST/CoastalProtection/WCVI folder. If you created your own workspace folder, then select it here. Click on the WCVI folder and click on set the main model workspace. This is the folder in which you will find the “_WaveModel_Outputs” (final outputs) folders after the model is run. 20. Specify the Label for Nearshore Waves and Erosion run. This string of text will be stripped of spaces and shortened to 10 characters. It will serve as the suffix to many of outputs. Type “Dune_2m” into the window. 21. Provide the Erosion Protection Excel Table. The model requires you to specify information about site informa- tion and habitat management actions. A sample Erosion Protection Excel Table will be supplied for you. Click and path to the InVEST/CoastalProtection/Input data folder. Double left-click on the file ErosionPro- tection_WCVI_BarkSound.xls. Click to make the selection. 22. Specify a Cross-Shore Profile. The model requires a text file of a smoothed bathymetric and topographic tran- sect. This can either be an output from the Profile Generator or a profile of your own. Click and path to the InVEST/CoastalProtection/Input data folder. Double left-click on the file In- VEST/CoastalProtection/WCVI/_ProfileGenerator_Outputs/Dune_2m/html_txt/CreatedProfile_Dune_2m.txt. Click to make the selection. 23. Select ‘(1) Yes, I have these values’ in answer to the question about whether you have wave height and period values. 24. Specify a Wave Height. Enter the wave height you wish to model. For this example, enter a value of “5” for this input. 25. Specify a Wave Period. Enter the wave period you wish to model. For this example, enter a value of “10” for this input. If ‘(2)No, Please compute these values from wind speed and fetch distance’ had been selected, which is only
7.6. Running the model 119 InVEST User Guide, Release 2.5.6
7.6. Running the model 120 InVEST User Guide, Release 2.5.6
appropriate for sheltered shorelines, you would have to provide a wind speed, a fetch length (distance from point of interest to adjacent land masses), and an average depth in along the transect of interest. 26. Specify a Storm Duration. Please provide the duration of the storm (strong waves and surge) you are modeling. The default value is 5 hours but you can change this value by typing directly into the text box. 27. Specify the Surge Elevation. The model requires the elevation of the peak surge relative to Mean Sea Level. The default value is 1 meter but you can change this value to better represent the storm conditions and tidal range at your site. You can change the value by typing directly into the text box. For this example, enter a value of “3” for this input. 28. Specify the Model Spatial Resolution. The default resolution is 1m. If you would like a coarser resolution to improve run time, you can increase this value by typing a larger value into the text box. 29. Compute Economic Valuation? Check this box if you would like to approximate the change in erosion damages due to a change in habitat from a management action. This requires the remainder of the fields to be populated. If economic valuation is not desired, do not check this box. 30. Specify the Longshore Extent. If you wish to compute economic valuation, you will have to provide a dis- tance along the shore where habitat, topo/bathy, forcing, habitat management actions, and property value are essentially uniform. This is a site specific parameter but 250m is the default value. 31. Specify the Property Value. If you wish to compute economic valuation, you will have to provide the property value of the nearshore land in your local currency per square meters. For this example enter a value of “12”. 32. Specify the Return Period of Storm. If you wish to compute economic valuation, you will have to provide the return period of the storm you are modeling (waves and surge). For example, if you are modeling surge and waves associated with the ‘100-year storm,’ enter a value of 100 here. For this example, enter a value of “25”. 33. Specify the Discount Rate. If you wish to compute economic valuation, you will have to provide a discount rate. The default value is 5% (0.05) but you are free to change this parameter if a different discount rate is more appropriate. 34. Specify the Time Horizon of Valuation. If you wish to compute economic valuation, you will have to provide the number of years into the future you would like to value the protective services of your habitat. 35. At this point the model dialog box is completed for a full run of the Nearshore Waves and Erosion portion of the Erosion Protection model. Click to start the model run. The model will begin to run and a show a progress window with progress information about each step in the analysis. Once the model finishes, the progress window will show all the completed steps and the amount of time that has elapsed during the model run.
7.6.3 Viewing output from the model
Upon successful completion of the model, two new folders called “_ProfileGenerator_Outputs” and “_Wave- Model_Outputs” will be created in each of the sub-models (Profile Generator and Nearshore Waves and Erosion) workspaces. They both contain a link to an html page that shows results of your run as well as various files that sup- plement the information on that html page. Output files are described in more detail in the Interpreting results section.
7.7 Interpreting results
7.7.1 Model outputs
The following is a short description of each of the outputs from the Erosion Protection model. Each of these output files is saved in the output workspace directory you specified:
7.7. Interpreting results 121 InVEST User Guide, Release 2.5.6
7.7. Interpreting results 122 InVEST User Guide, Release 2.5.6
7.7. Interpreting results 123 InVEST User Guide, Release 2.5.6
_ProfileGenerator_Outputs
This folder contains a sub-folder whose name is the “suffix label” you specified in this model’s interface. It contains two sub-folders: html_txt and maps. html_txt
This folder contains two webpage links, figures used in the webpages, and three text files. + profile.html: This HTML file contains information summarizing the location of your site, as well as the information you entered in the model’s interface and Excel input file. This output also contains figures showing the bathymetry profile created and/or smoothed by the Profile Generator Model, with close ups of the backshore area, when applicable. Also, if you have uploaded a folder of natural habitats and* used the Profile Generator Model to cut a cross-shore transect for you from a DEM file, a table and figure are presented that indicate the X-coordinates of the beginning and end of where each natural habitat exists along the transect. • fetchwindwave.html: This HTML file contains figures showing wind and fetch roses. It also contains infor- mation on fetch distances computed by the model, if you chose this option. There are also tables showing the average values of the maximum, as well as the top 10% and 25% wind speed and wave height extracted from the WW3 gage point closest to your site, if you uploaded that file. Finally, if you had the model compute fetch distances for you and uploaded WW3 data, this page also contains estimates of wind-generated wave height for each of the 16 equidistant sectors that make a full compass circle. • FetchDistances_[suffix].txt: This text file contains information on fetch distances computed by the model. It has two columns. The first column shows the 16 directional sectors angles, and the second column has fetch distances associated with these sectors. • BathyProfile_[suffix].txt: This text file is the smoothed bathymetric profile produced by the Profile Generator. It only contains values of water depths below MLLW (or the vertical datum of your bathy or topo/bathy DEM).
7.7. Interpreting results 124 InVEST User Guide, Release 2.5.6
The first column consists of X-values with X=0 at the shoreline, and the second column corresponds to depths values at the various cross-shore X distances. • CreateProfile_[suffix].txt: This text file is the smoothed bathymetric and topographic profile produced by the Profile Generator. It differs from “BathyProfile_label.txt” because it has the backshore information. This back- shore information was either provided by information in the Erosion Protection Excel Table or extracted from your DEM if you provided a seamless DEM. We recommend that you use this profile as input in the Erosion Protection model. • ProfileCutGIS_[suffix].txt: This text file is the un-smoothed and un-processed raw profile that was cut by the model, if you chose that option, before smoothing and/or the addition of backshore information. This file is useful if you want to see the quality of the GIS DEM data that you uploaded. If you have a good quality DEM layer that contains a high resolution representation of your area, this text file can also be useful and input in the wave model, as long as it is smoothed. maps
• Fetch_Vectors.shp: This polyline shapefile depicts the remaining fetch radials found in the seascape after being intersected with the Land Polygon input (landscape) you provided . The GIS starts with 144 vectors in total, at 2.5 degree increments, and erases all radials that overlap with the landscape. • Fetch_Distances.shp: This polyline shapefile summarizes fetch distances for Land Point input over 16 directions that you specified. • Profile_Pts.shp: This point shapefile represents the cross-shore transect that was cut by the GIS. Its attribute table contains depth information from both the raw and smoothed profiles. • Profile_Pts_Hab.shp: This point shapefile represents the cross-shore transect that was cut by the GIS and then intersected with the habitat layers you provided. In the attribute table, columns for each of the six possible habitats are included. A value of “1” means a particular habitat is present at a point along the transect, while a “0” means it is not found.
_WaveModel_Outputs
This folder contains two useful outputs from the Nearshore Waves and Erosion model: • OutputWaveModel_[suffix].html: This HTML file summarizes the information you entered as input in the model, including wave forcing and habitat management actions, and describes the outputs. It contains a fig- ure depicting profiles of wave height (before and after habitat management action), as well as percent of wave attenuation and the location of your natural habitats along your bathymetry. It also provides a figure showing a profile of erosion or hourly rate of bed scour in your backshore area before and after management action. If valuation was selected, a table summarizing the value of your natural habitats before and after management action is presented. • WaveHeight_[suffix].txt: This text file contains three columns showing distance from the shoreline and profiles of wave height over your bathymetry profile, before (second column) and after (third column) your management action. • WaveHeightAfter_[suffix].txt: This text file contains two columns showing distance from the shoreline and profiles of wave height over your bathymetry profile, before after your management action. • WaveHeightBefore_[suffix].txt: This text file contains two columns showing distance from the shoreline and profiles of wave height over your bathymetry profile, before your management action.
7.7. Interpreting results 125 InVEST User Guide, Release 2.5.6
7.7.2 Parameter log
Each time the module is run a text file will appear in the workspace folder. The file will list the parameter values for that run and be named according to the service and the date and time.
7.8 References
Armono, and K. Hall. (2003). Laboratory study of wave transmission on artificial reefs. Proc. Canadian Coastal Eng. Conf., Kingston, Canada Apotsos, A., B. Raubenheimer, S. Elgar, and R.T. Guza. (2008). Testing and calibrating parametric wave transforma- tion models on natural beaches, Coast. Eng., 55. Alsina, J. M. and T. E. Baldock (2007). “Improved representation of breaking wave energy dissipation in parametric wave transformation models.” Coastal Eng. 54(10). Battjes, J.A. and M.J.F. Stive (1985). Calibration and verification of a dissipation model for random breaking waves, J. Geophys. Res., 90(C5). Bradley, K., and C. Houser (2009), Relative velocity of seagrass blades: Implications for wave attenuation in low- energy environments, J. Geophys. Res., 114. Burger, B. (2005). Wave Attenuation in Mangrove Forests, A Master’s thesis publication, Delft U. of Technology, Civil Engineering and Geosciences. Cannon, M.G., J.M. Phelan, and M.A. Passaro. (1995) Procedural guidelines for estimating residential and business structure value for use in flood damage estimations. IWR Report 95-R-9, USACE. Water Resources Support Center, Alexandria, Virginia Dean, R. G. and C. J. Bender (2006) “Static Wave Setup With Emphasis on Damping Effects by Vegetation and Bottom Friction”, Coastal Eng. 13. Dean, R.G. and R.A. Dalrymple. (2002) Coastal Processes with Engineering Applications. Cambridge University Press. 475pp. de Vos, J.W. (2004). Wave attenuation in mangrove wetlands: Red River Delta, Vietnam. MSc thesis. Delft University of Technology Gourlay, M.R. (1996a).Wave set-up on coral reefs. 1. Set-up and wave-generated flow on an idealised two dimensional reef. J. Coastal Eng. 27. Gourlay, M.R. (1996b).Wave set-up on coral reefs. 2.Wave set-up on reefs with various profiles. J. Coastal Eng. 28. Gourlay, M.R. (1997). Wave set-up on coral reefs: some practical applications. Proc. Pacific Coasts and Ports, Christchurch, 2, 959–964. Keddy, P. A. (1982). Quantifying within-lake gradients of wave energy: Interrelationships of wave energy, substrate particle size, and shoreline plants in Axe Lake, Ontario. Aquatic Botany 14, 41-58. Kobayashi, N., A. W. Raichle, and T. Asano (1993), Wave attenuation by vegetation, J. Waterw. Port, Coastal Ocean Eng., 119 Koch, E.W., L.P. Sanford, S.N. Chen, D.J. Shafer and J.M. Smith (2006). Waves in seagrass systems: review and technical recommendations. US Army Corps of Engineers Technical Report. Engineer Research and Development Center, ERDC TR-06-15, 82 p. Komar, P.D. (1998) Beach Processes and Sedimentation, Prentice Hall, Upper Saddle River, N.J., 543pp. Kriebel, D. L., and R.G. Dean. (1993). Convolution method for time dependent beach-profile response. J. Waterw., Port, Coastal, Ocean Eng., 119(2) Mazda, Y., E. Wolanski, B. King, A. Sase, D. Ohtsuka, and M. Magi (1997). Drag force due to vegetation in mangrove swamps. Mangroves and Salt Marshes 1:193–99.
7.8. References 126 InVEST User Guide, Release 2.5.6
McLachlan, A. and Dorvlo, A. (2005). Global patterns in sandy beach macrobenthic communities. Journal of Coastal Research 21, 674-687. Mendez, F. J., and I. J. Losada (2004), An empirical model to estimate the propagation of random breaking and nonbreaking waves over vegetation fields, Coastal Eng., 51 Narayan, S. (2009). The Effectiveness of Mangroves in Attenuating Cyclone – induced Waves, Master’s thesis, Delft U. of Tech., Civil Eng. and Geosciences Stockdon, H.F., R.A. Holman, P.A. Howd, and A.H. Sallenger. (2006). Empirical parameterization of setup, swash, and runup. Coastal Engineering, 53 Thornton, E. and Guza, R.T. (1983). Transformation of Wave Height Distribution. Journal of Geophysical Research 88(C10) Tolman, H.L. (2009). User manual and system documentation of WAVEWATCH III version 3.14, Technical Note, U. S. Department of Commerce Nat. Oceanic and Atmosph. Admin., Nat. Weather Service, Nat. Centers for Environmental Pred., Camp Springs, MD. U.S. Army Corps of Engineers (USACE) (2002). U.S. Army Corps of Engineers Coastal Engineering Manual (CEM) EM 1110-2-1100 Vicksburg, Mississippi. Van der Meer, J.W., Briganti, R., Zanuttigh, B. and Wang, B. (2005). Wave transmission and reflection at low crested structures: design formulae, oblique wave attack and spectral change, Coast. Eng., 52. Whitehouse, R., Soulsby, R.R., Roberts, W., Mitchener, H. (2000). Dynamics of Estuarine Muds. H. R. Wallingford, UK Wiegel, R.L. 1964 Oceaographical Engineering, Prentice-Hall, Englewood Cliffs, NJ.
7.9 Appendix A
7.9.1 Beach Survey with “Emery Boards”
(Adapted from Beach Profiling with “Emery Boards” and Measuring Sand Grain Size, 2005, Florida Center for Instructional Technology, University of South Florida) The simplest technique to measuring a beach profile is known as the “Emery board” method, developed by a famous coastal scientist named K.O. Emery. As depicted in Figure 1 the apparatus consists of two stakes connected by a rope of known length (5m or 10m). This length sets the measurement interval for individual data points along the profile. Each stake has a measurement scale which runs from 0 at the top, down to the bottom of the stake. It is recommended to use Metric units. This approach may seem simple, but it provides reasonably accurate measurements of beach profiles. It also has the advantages of light, inexpensive, equipment, which can be easily carried to distant survey sites, for very rapid surveys. The technique of measuring sand size will be conducted in the field with the use of sand gauge charts. These are small, credit-card sized, plastic charts with calibrated samples of sieved sand mounted on the face. By using a hand-lens and sand gauge chart, it is possible to compare samples from the beach with calibrated samples on the chart for an estimate of size range. Sand gauge charts are available from a number of vendors. One such distributor is ASC Scientific.
7.9.2 Materials
To build a set of “Emery boards”, all that is needed are two pieces of wood of equal length and a rope of known length. (Boards slightly smaller than observers will work well (~1.6m).) Tie a loop in each end of the rope, which can easily slide up and down the two boards. Measuring down from the top of each board, use a marker and a ruler to draw and label the “graduations” (marks of equal length). An appropriate graduation interval is every two centimeters. Additionally, one can attach a small level to the rope to help ensure it is horizontal (for example).
7.9. Appendix A 127 InVEST User Guide, Release 2.5.6
Figure 7.20: Illustration of the Emery Board technique
7.9.3 Method
At the very minimum, two people are necessary to conduct a survey, but three are preferable. Team members should separate themselves into a “seaward surveyor”, a “landward surveyor”, a “geotechnical engineer” and a “data recorder”. The “seaward surveyor” is responsible for holding the seaward board and ensuring that the rope is level between the two boards (by sliding the loop up or down) when fully extended. The “landward surveyor” is responsible for holding the landward board, sighting over the seaward board to the horizon, and shouting out the measurement (cm down from the top of the landward board) to the “data recorder”. The “geotechnical engineer” is responsible for moving with the “seaward surveyor” to collect a sand sample, and identify it using the hand lens on the basis of its size comparison to the sand gage chart. The “data recorder” should keep organized notes of each measurement including horizontal distance (x), measurement of change in elevation (a), cumulative change in elevation of all measurements, and sand size at each location. Starting at the landward extent of the survey region (baseline), cross-shore data points of elevation and sand size are collected at the sampling interval determined by the length of the rope (distance between the two boards at full extension). Collect at least 5 cross shore data points. Collect more than 5 cross shore data points if the beach is wide. If the beach is sloping downward toward the sea, the observer sights across the top of the seaward board to the level of the horizon, and determines the distance (A1) from the top of the landward board to the sightline in the following figure (or distance (a) in Figure 1). If the beach is locally sloping upward in the offshore direction, then (A2) is measured on the seaward board and the sighting is with the horizon over the top of the landward board (next figure). If horizon cannot be found on landward side, then observer on landward aligns his/her eye with pointer (pen or other thin sharp object) adjusted and held by observer on seaward side and horizon to form a horizontal line. Observer on seaward side then reads distance A2, which should be recorded as negative to indicate upward slope. In addition, the “data recorder” should make notes of the time of observations and such things such as presence/absence and type of beach debris (kelp, wood etc.). Also, the “data recorder” should take note of the maximum landward extent of these debris if they were freshly deposited, as an indication of position of high tide. High tide location can also be guessed by looking for position of wet/dry sand barrier. If the team has a portable GPS unit, the “data recorder” should note the coordinate of this high water mark, or if there are repeated measurements at the same site, the “data recorder” should evaluate its distance from known landmark. Finally, the “data recorder” should make note of position (GPS or meters) of position of landward board during first measurement, of seaward board after last measurement, and position
7.9. Appendix A 128 InVEST User Guide, Release 2.5.6
Figure 7.21: Same as previous figure. Find distance A1 from top of board to eye such that eye, top of board 2 and horizon are aligned. Line must be horizontal.
Figure 7.22: Same as two previous figures. Find distance A2 from top of board to pointer such that eye at top of board 1, pointer and horizon are aligned. Line must be horizontal.
7.9. Appendix A 129 InVEST User Guide, Release 2.5.6 of water level.
7.9.4 Recording and Processing Data
Assuming that the rope is 10m long, an example log looks as follow, where positive values are A1 measurements (sloping down), and negative values are A2 measurements (sloping up):
Based on these values, a beach profile can be constructed by performing the following operations:
Measured values are in column 1, and cumulative distance between measurements is in Column 2 (assuming rope is 10m long). In Column 3 we estimate 1/Slope, using DX=length of rope=10m. For example, slope of 1st measurement is 1/2. In Column 4, we estimate beach profile, assuming that zero is located at point where first measurement is taken. In Column 5 we estimate beach profile again, assuming that zero is last point measured. This last column is used to plot profile of beach as function of X, as shown in the following figure. Finally, if repeated measurements are made at the same time, it is recommended to continuously log time of mea- surement, and positions of board at beginning and end of measurement, as well as high water mark. These should be indicated on beach profile, if possible. Also, by looking at tide chart, it is possible to estimate high water level during period of measurement, and use this info to convert beach profile values accordingly.
7.9. Appendix A 130 InVEST User Guide, Release 2.5.6
Figure 7.23: Example beach profile measured with Emery Board. Zero is last point measured.
7.9. Appendix A 131 CHAPTER EIGHT
MARINE FISH AQUACULTURE
8.1 Summary
Supporting the production of aquacultured fish and shellfish is an important service provided by coastal and marine environments. Because salmon is one of the two most important finfish in aquaculture worldwide, the current version of the InVEST aquaculture model analyzes the volume and economic value of Atlantic salmon (Salmo salar) grown in netpen aquaculture facilities based on farming practices, water temperature, and economic factors. Inputs for the present model include farm location, management practices at the facilities, water temperature, economic data for valuation, and the time period over which results are of interest. This model is best used to evaluate how human ac- tivities (e.g., the addition or removal of farms or changes in harvest management practices) and climate change (e.g., change in sea surface temperature) may affect the production and economic value of aquacultured Atlantic salmon. Limitations of the model include assumptions that harvest practices, prices, and costs of production of aquacultured fish are constant for the selected time period. Additionally, risk of disease outbreaks and variability between individual salmon within a farm are not included in the model. Future releases of this model will include the following features: 1) guidance for modifying the Atlantic salmon model for other aquacultured marine fish, 2) quantification of wastes produced at aquaculture facilities, 3) a separate module for quantifying volume, economic value, filtration and pro- duction of wastes of aquacultured shellfish (e.g., oyster, shrimp), and 4) a sub-module to evaluate impacts of parasitic sea-lice on farmed Atlantic salmon. This is a “Tier 1” model.
8.2 Introduction
Human demand for protein from the ocean has rapidly increased and is projected to continue to do so in coming decades (Delgado et al. 2003, Halwart et al. 2007, Soto et al. 2008). In recent years, the scales, previously tilted towards provisioning of protein from capture fisheries, have shifted toward aquaculture. In particular, finfish aquacul- ture, primarily for Atlantic salmon, has intensified in coastal areas over the past two decades (FAO 2004, Goldburg and Naylor 2004, Naylor and Burke 2005). In 2002, farmed salmon production, over 90% of which was for Atlantic salmon, was 68% higher than the volume of wild capture (FAO 2004). Atlantic salmon farming, conducted in float- ing netpens in low energy, nearshore areas, is a well-established, consolidated industry that operates in the temperate waters of Norway, Chile, the United Kingdom and Canada. Commercial operations for Atlantic salmon use the marine environment to produce a valuable commodity, which gen- erates revenue and is a source of employment. Yet salmon farming is controversial due to potentially adverse impacts to marine ecosystems and, thereby, people who derive their livelihoods from those ecosystems (e.g., commercial fish- ermen, tourism operators). Concerns about the effects of Atlantic salmon aquaculture on the marine ecosystem involve debate about the impacts of emission of dissolved and solid wastes to water quality and living habitats, degradation of water quality due to use of antibiotics, mixing and competition of escaped farmed salmon with endemic species (e.g., Pacific salmon), increased risk of parasitism and disease, and depletion of forage fish resources harvested from other ecosystems for use as Atlantic salmon feed.
132 InVEST User Guide, Release 2.5.6
Regulations for the Atlantic salmon aquaculture industry vary regionally, from the most stringent requirements for locating and operating facilities in Norwegian waters, to fewer constraints for farms in Chilean waters. For all oper- ations, there are regulatory limits on where and how aquaculture can be conducted and requirements for monitoring and regulating the amount of waste generated at different facilities, and in some cases, mitigation requirements. Weighing the economic benefits of Atlantic salmon aquaculture against the environmental costs involves quantifying both. The InVEST model presented here does the former by quantifying the volume and economic value of the commodity, and will include the latter in the next model release (May 2011). Intermediate outputs will include dissolved and particulate wastes generated as a byproduct of Atlantic salmon production. These outputs will be available for use in other InVEST models (e.g., water quality, habitat quality, fisheries) to assess impacts of Atlantic salmon aquaculture on other coastal and marine environmental services. With the full suite of model outputs, InVEST users will be able explore how different spatial configuration of Atlantic salmon farms in their region affects other ecosystem benefits and alleviates or exacerbates tradeoffs between economic benefits and downstream environmental costs.
8.3 The Model
The model is designed to address how the production and economic value of farmed Atlantic salmon at individual aquaculture facilities and across a user-defined study area change depending on farm operations and changes in water temperature. Temporal shifts in price, costs or harvest management practices are not dynamically modeled, but can be represented by running the model sequentially, where each run uses different information on prices, costs and farm operations. The risk of disease outbreaks and variability between individual salmon within a farm are not included in the model. The model will yield the most accurate outputs when parameterized with site-specific temperature and farm operations data. If site-specific data are unavailable, the provided ranges of default values can be used to yield first approximations of results (see Data needs section). The model is run simultaneously for all Atlantic salmon farms identified by the user. Each farm can have a user-defined set of operations and management practices. The volume of fish produced on a farm depends on water temperature (which affects growth), the number of fish on the farm, the target harvest weight range, and the mortality rate. Fish growth is modeled on a daily time-step until the fish reach the target harvest weight range, after which they are harvested. After a user-defined fallowing period, the farm is restocked and this initiates the next production cycle. Production cycles continue for each farm until the end of the time period of interest (e.g., 2 years, 10 years). Outputs include the harvested weight of fish and net revenue per cycle for each individual farm. In addition, the model yields a map of the total harvested weight, total net revenue, and net present value over the time period of interest.
8.3.1 How it works
The model runs on a vector GIS dataset that maps individual aquaculture facilities for Atlantic salmon that are actively farmed over a user-defined time period. The map can be based on current farming (the “status quo” or “baseline” scenario), or on scenarios of projected expansion or contraction of the industry or on projected changes in water temperature. In each farm we model the production of fish in three steps. (1) We model the growth of individual fish to harvest weight. (2) We calculate the total weight of fish produced in each farm as the number of fish remaining at harvest, multiplied by their harvested weight, less the weight removed during processing (gutting, etc.) and the weight of fish lost to natural mortality. (3) Lastly, all the fish in a farm are harvested at the same time, and the farm is restocked after a user-defined fallowing period. Valuation of processed harvest is an optional fourth step in the model.
Growth of the individual fish to harvest weight
Atlantic salmon weight (kg) is modeled from size at outplanting to target harvest weight. Weight is a function of growth rate and temperature (Stigebrandt 1999). Outplanting occurs when Atlantic salmon have been reared beyond
8.3. The Model 133 InVEST User Guide, Release 2.5.6
their freshwater life stages. The model runs on a daily time step because the next version of the model (to be released in May 2011) will quantify waste products from aquaculture farms for use as inputs into the Marine InVEST water quality model. Fine resolution temporal data are more appropriate for the seasonal evaluation of environmental impacts (e.g., seasonal eutrophication).
Weight Wt at time t (day), in year y, and on farm f is modeled as:
β Tt−1,f τ Wt,y,f = (αWt−1,y,f · e ) + Wt−1,y,f (8.1)
1-b -1 where α (g day ) and b (non-dimensional) are growth parameters, Tt,f is daily water temperature (C) at farm f, and τ (0.08 C-1) is a fixed scalar that represents the doubling of biochemical rates in fish when temperature increases by 8-9 C. Daily water temperatures can be interpolated from monthly or seasonal temperatures. The growing cycle for each farm begins on the user-defined date of outplanting (t = 0). The outplanting date is used to index where in the temperature time series to begin. The initial weight of the outplanted fish for each farm is user-defined. An individual Atlantic salmon grows until it reaches its target harvest weight range, which is defined by the user as a target harvest weight.
Total weight of fish produced per farm
To calculate the total weight of fish produced for each farm, we assume that all fish on a farm are homogenous and ignore variability in individual fish growth. This assumption, though of course incorrect, is not likely to affect the results significantly because 1) netpens are stocked so as to avoid effects of density dependence and 2) aquaculturists outplant fish of the same weight to netpens for ease of feeding and processing. We also assume that when fish reach a certain size, all fish on the farm are harvested. In practice, farms consist of several individual netpens, which may or may not be harvested simultaneously. If a user has information about how outplanting dates and harvest practices vary between netpens on a farm, the user can define each netpen as an individual “farm.” The total weight of processed fish TPW on farm f in harvest cycle c:
−M·(th−t0) TPWf,c = Wth,h,f · d · nf e (8.2)
where Wth,h,f is the weight at date of harvest th, y on farm f from Equation (8.1), d is the processing scalar which is the fraction of the fish in the farm that remains after processing (e.g., weight of headed/gutted or filleted fish relative −M·(t −to) to harvest weight), nf is the user-defined number of fish on farm f, and e h is the daily natural mortality rate M experienced on the farm from the date of outplanting (t0) to date of harvest (th).
Restocking
The previous 2 steps describe how fish growth is mdoeled for one production cycle. However, the user may want to evaluate production of fish over a series of production cycles. The primary decision to be made when modeling multiple harvest cycles is if (and if so, how long) a farm will be left to lie fallow after harvest and before the next production cycle begins (initiated by outplanting). If used, fallowing periods are considered hard constraints in the model such that a farm cannot be restocked with fish until it has lain fallow for the user-defined number of days. This is because fallowing periods are often used to meet regulatory requirements, which can be tied to permitting, and thus provide incentive for compliance. Once fish are harvested from a farm and after the user-defined fallowing period, new fish are outplanted to the farm. The model estimates the harvested weight of Atlantic salmon for each farm in each production cycle. The total harvested weight for each farm over the time span of the entire model run is the sum of the harvested weights for each production cycle.
8.3. The Model 134 InVEST User Guide, Release 2.5.6
Valuation of processed fish (optional)
The aquaculture model also estimates the value of that harvest for each farm in terms of net revenue and net present value (NPV) of the harvest in each cycle. The net revenue is the harvest weight for each cycle multiplied by market price, where costs are accounted for as a fraction of the market price for the processed fish. Fixed and variable costs, including costs of freshwater rearing, feed, and processing will be more explicitly accounted for in the next iteration of this model. The NPV of the processed fish on a farm in a given cycle is the discounted net revenue such that: 1 NPVf,c = TPWf,c[p(1 − C)] · (8.3) (1 + r)t
where TPWf,c is the total weight of processed fish on farm f in harvest cycle c, p is the market price per unit weight of processed fish, C is the fraction of p that is attributable to costs, r 1 is the daily market discount rate, and t is the number of days since the beginning of the model run.
Note: The beginning of the model run is the initial outplanting date for the very first farm (of all the farms in the study area) to receive fish. Thus, the net revenue for each farm in each harvest cycle is discounted by the number of days since the very first farm was initially stocked. The total NPV for each farm over the duration of the model run is the discounted net revenue from each harvest cycle summed over all harvest cycles c.
The discount rate reflects society’s preference for immediate benefits over future benefits (e.g., would you rather receive $10 today or $10 five years from now?). The default annual discount rate is 7% per year, which is one of the rates recommended by the U.S. government for evaluation of environmental projects (the other is 3%). However, this rate can be set to reflect local conditions or can be set to 0%.
Uncertainty analysis (optional)
Optionally, if the fish growth parameters are not known with certainty, the model can perform uncertainty analy- sis. This uncertainty analysis is done via a Monte Carlo simulation. In this simulation, the growth parameters are repeatedly sampled from a given normal distribution, and the model is run for each random sampling. The results for each run of the simulation (harvested weight, net present value, and number of completed cycles per farm) are collected and then analyzed. Uncertainty results are output in two ways: first, the model outputs numerical results, displaying the mean and the standard deviation for all results across all runs. Second, the model creates histograms to help visualize the relative probability of different outcomes.
8.4 Limitations and simplifications
Limitations of the model include assumptions that harvest practices, prices, and costs of production of aquacultured fish are constant over the selected time period. Additionally, risk of disease outbreaks and variability between individ- ual salmon within a farm are not included in the model. The current model operates at a daily time step (requiring daily temperature data), but future iterations will allow for monthly or yearly temperature inputs. Uncertainty in input data is currently supported only for fish growth parameters. There is currently no support for uncertainty in input data such as water temperature.
1 The daily discount rate is computed as the annual discount rate divided by 365. For an annual discount rate of 7%, the daily discount rate is 0.00019178.
8.4. Limitations and simplifications 135 InVEST User Guide, Release 2.5.6
8.5 Data needs
8.5.1 Data sources
Here we outline the specific data and inputs used by the model and identify potential data sources and default values. Four data layers are required, and one is optional (but required for valuation). 1. Workspace Location (required). Users are required to specify a workspace folder path. It is recommended that the user create a new folder for each run of the model. For example, by creating a folder called “runBC” within the “Aquaculture” folder, the model will create “intermediate” and “output” folders within this “runBC” workspace. The “intermediate” folder will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “output” folder.: Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\Aquaculture\runBC
2. Finfish Farm Location (required). A GIS polygon or point dataset, with a latitude and longitude value and a numerical identifier for each farm.: Names: File can be named anything, but no spaces in the name File type: polygon shapefile or .gdb Rows: each row is a specific netpen or entire aquaculture farm Columns: columns contain attributes about each netpen (area, location, etc.). Sample data set: \InVEST\Aquaculture\Input\Finfish_Netpens.shp
Note: The user must ensure that one field contains unique integers. This field name (i.e. “FarmID” in the sample data) must be chosen by the user for input #3 as the “farm identifier name”.
Note: The model checks to ensure that the finfish farm location shapefile is projected in meters. If it is not, the user must re-project it before running the model.
3. Farm Identifier Name (required). The name of a column heading used to identify each farm and link the spatial information from the GIS features (input #2) to subsequent table input data (farm operation and daily water temperature at farm tables, inputs # 6-7). Additionally, the numbers underneath this farm identifier name must be unique integers for all the inputs (#2, 6, & 7).: Names: A string of text identifying a column in the Finfish Farm Location shapefile’s attribute table File type: Drop-down option Sample: FarmID
4. Fish growth parameters (required, defaults provided). Default a (0.038 g/day) and b (0.6667 dimensionless units) are included for Atlantic salmon, but can be adjusted by the user as needed. If the user chooses to adjust these parameters, we recommend using them in the simple growth model (Equation (8.1)) to determine if the time taken for a fish to reach a target harvest weight typical for the region of interest is accurate.: Names: A numeric text string (floating point number) File type: text string (direct input to the ArcGIS interface) Sample (default): 0.038 for a / 0.6667 for b
5. Uncertainty analysis data (optional). These parameters are required only if uncertainty analysis is desired. Users must provide three numbers directly through the tool interface.: • Standard deviation for fish growth parameter a. This represents uncertainty in the estimate for the value of a. • Standard deviation for fish growth parameter b. This represents uncertainty in the estimate for the value of b.
8.5. Data needs 136 InVEST User Guide, Release 2.5.6
• Number of Monte Carlo simulation runs. This controls the number of times that the parameters are sampled and the model is run, as part of a Monte Carlo simulation. A larger number will increase the reliability of results, but will also increase the running time of the model. Monte Carlo simulations typically involve about 1000 runs. 6. Daily Water Temperature at Farm Table (required). Users must provide a time series of daily water tempera- ture (C) for each farm in data input #1. When daily temperatures are not available, users can interpolate seasonal or monthly temperatures to a daily resolution. Water temperatures collected at existing aquaculture facilities are preferable, but if unavailable, users can consult online sources such as NOAA’s 4 km AVHRR Pathfinder Data and Canada’s Department of Fisheries and Oceans Oceanographic Database. The most appropriate temperatures to use are those from the upper portion of the water column, which are the temperatures experienced by the fish in the netpens.: Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Office 2007 or newer) Rows: There are 365 rows (rows 6-370), each corresponding to a day of the year. Columns: The first two columns contain the number for that year (1-365) and day-month. Sample: \InVEST\Aquaculture\Input\Temp_Daily.xls\WCVI$
Note: For clarification on rows, please refer to the sample temperature dataset in the InVEST package (Temp_Daily.xls).
Note: Column “C” and then all others to its right contain daily temperature data for a specific farm, where the numbers found in row 5 must correspond to the numbers underneath the farm identifier name found in input #2’s attribute table.
7. Farm Operations Table (required). A table of general and farm-specific operations parameters. Please refer to the sample data table for reference to ensure correct incorporation of data in the model. If you would like to use your own dataset, you can modify values for farm operations (applied to all farms) and/or add new farms (beginning with row 32). However, do not modify the location of cells in this template. If for example, you choose to run the model for three farms only, they should be listed in rows 10, 11 and 12 (farms 1, 2, and 3, respectively). Several default values that are applicable to Atlantic salmon farming in British Columbia are also included in the sample data table. The majority of these values can be found by talking to aquaculturists in the study area or through regional industry reports from major aquaculture companies (e.g. Panfish, Fjord Seafood, Cermaq, Marine Harvest, Mainstream Canada, and Grieg). The General Operation Parameters of the input table includes the following inputs that apply to all farms: + Fraction of the fish weight (in the farm) remaining after processing (e.g., weight of headed/gutted fish relative to harvest weight) + Natural mortality rate on the farm (daily) + Duration of simulation (in years) The Farm-Specific Operation Parameters of the input table includes the following inputs: • Rows: Each row in this table (table begins at row #10) contains the input data for a specific farm. • Columns: Each column contains values and should be named as follows: – Farm #: a series of consecutive integers (beginning with “1” in row 10) that identifies each farm and must correspond to the unique integers underneath the farm identifier name found in input #2’s attribute table. – Weight of fish at start (kg): this is the weight of fish when they are outplanted, which occurs when Atlantic salmon have been reared beyond their freshwater life stages. – Target weight of fish at harvest (kg) – Number of fish in farm (absolute) – Start day for growing (Julian day of the year): this is the date of the initial outplanting at the start of the model run. Outplanting date will differ in subsequent cycles depending on lengths of growth and fallowing periods.
8.5. Data needs 137 InVEST User Guide, Release 2.5.6
– Length of fallowing period (number of days): if there is no fallowing period, set the values in this column to “0”. Table Names: File can be named anything, but no spaces in the name File type: *.xls or .xlsx (if user has MS Office 2007 or newer) Sample: \InVEST\Aquaculture\Input\Farm_Operations.xls\WCVI$
8. Run Valuation? (optional). By checking this box, users request valuation analysis. 9. Valuation parameters (required for valuation, defaults provided).: Names: A numeric text string (positive integer or floating point number) File type: text string (direct input to the ArcGIS interface) Sample (default): a. Market price per kilogram of processed fish. Default value is 2.25 $/kilogram (Urner-Berry monthly fresh sheet reports on price of farmed Atlantic salmon) b. Fraction of market price that accounts for costs rather than profit. Default value is 0.3 (30%). c. Daily market discount rate. We use a 7% annual discount rate, adjusted to a daily rate of 0.000192 for 0.0192% (7%/365 days).
Note: If you change the market price per kilogram, you should also change the fraction of market price that accounts for costs to reflect costs in your particular system.
8.6 Running the model
Note: The word ‘path’ means to navigate or drill down into a folder structure using the Open Folder dialog window that is used to select GIS layers or Excel worksheets for model input data or parameters.
8.6.1 Exploring the workspace and input folders
These folders will hold all input, intermediate and output data for the model. As with all folders for ArcGIS, these folder names must not contain any spaces or symbols. See the sample data for an example.
Exploring a project workspace and input data folder
The /InVEST/Aquaculture folder holds the main working folder for the model and all other associated folders. Within the Aquaculture folder there will be a subfolder named ‘Input’. This folder holds most of the GIS and tabular data needed to setup and run the model. The following image shows the sample folder structure and accompanying GIS data. We recommend using this folder structure as a guide to organize your workspaces and data. Refer the following screenshots below for examples of folder structure and data organization.
8.6.2 Creating a run of the model
The following example describes how to set up the Aquaculture model using the sample data provided with the InVEST download. We expect users to have location-specific data to use in place of the sample data. These instructions provide
8.6. Running the model 138 InVEST User Guide, Release 2.5.6
only a guideline on how to specify to ArcGIS the various types of data needed and do not represent any site-specific model parameters. See the Data needs section for a more complete description of the data specified below. 1. Click the plus symbol next to the InVEST toolbox.
2. Expand the Marine toolset and click on the Finfish Aquaculture script to open the model. 3. Specify the Workspace. Open the InVEST workspace. If you created your own workspace folder (Step 1), then select it here. Select the Aquaculture folder and click to set the main model workspace. This is the folder in which you will find the intermediate and final outputs when the model is run. 4. Specify the Finfish Farm Location. This represents the geographic area over which the model will be run. This example refers to Finfish_Netpens.shp supplied in the sample data. Open the InVEST/Aquaculture/Input data folder.
5. Specify the Farm ID Field. The model requires the selection of one attribute heading from the Finfish Farm Location shapefile that contains a unique farm ID. For this example, select the ‘FarmID’ directly from the drop-down list. 6. Specify the Fish Growth Parameters (a) and (b). These values are the growth parameters required by the model. Default values of 0.038 and 0.6667 (appropriate for Atlantic salmon only) are supplied for you. You can type directly into the text box to specify different values.
8.6. Running the model 139 InVEST User Guide, Release 2.5.6
7. Specify the Temperature Data. The model requires an Excel table of daily time series of temperature data. Open the InVEST/Aquaculture/Input data folder. Double left-click on Temp_Daily.xls and select the worksheet WCVI$. Click to make the selection.
Note: ArcMap and the model will not recognize the Excel sheet as valid data if it is added to the Data View. It is best to add Excel data directly to the model using the Open and Add buttons and navigating to the data.
8. Specify the Farm Operations Data. The model requires an Excel table of farm-specific operation data. Open the InVEST/Aquaculture/Input data folder, double left-click Farm_Operations.xls and select WCVI$. Click to add the farm operations table. 9. Choose whether to run the economic valuation. Users can check the Run Valuation to conduct an economic valuation analysis. 10. Specify the market price of processed fish ($/per kilogram) (Optional). This optional parameter is the market price for a specific processed fish species. The default is given as 2.25 for Atlantic salmon. Users can enter a different value by typing directly into the text box. 11. Specify the fraction of market prices attributable to costs (Optional). This optional parameter is the fraction of market price attributable to costs. The default is given as 0.3. Users can enter a different value by typing directly into the text box. 12. Specify the daily market discount rate (Optional). This optional parameter is the discount rate for a type of fish. The default is given as 0.000192 (0.0192%). Users can enter a different value by typing directly into the text box. 13. At this point the model dialog box is ready for a complete run of the Finfish Aquaculture model. Click to start the model. The model will begin to run and a show a progress window with progress information about each step in the analysis. Once the model finishes, the progress window will show all the completed steps and the amount of time necessary to complete the model run.
8.6.3 Viewing output from the model
Upon successful completion of the model, you will see new folders in your Workspace called “intermediate” and “Output”. The Output folder, in particular, may contain several types of spatial data, each of which are described the Interpreting results section.
8.6. Running the model 140 InVEST User Guide, Release 2.5.6
8.6. Running the model 141 InVEST User Guide, Release 2.5.6
The Results.html file located in InVEST\Aquaculture\Output can help you to interpret the model results in terms of fish production and the economic valuation. You can view the output spatial data in ArcMap (Finfish_Harvest.shp shapefile) using the Add Data button .
You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting “Properties”, and then “Symbology”. There are many options here to change the way the data appear in the map. You can also view the attribute data of output files by right clicking on a layer and selecting “Open Attribute Table”.
8.7 Interpreting results
8.7.1 Model outputs
The following is a short description of each of the outputs from the Aquaculture tool. Each of these output files is automatically saved in the “Output” folder that is saved within the user-specified workspace directory: Final results are found in the output folder of the workspace for this model. The model produces two main output files: • Output\Finfish_Harvest.shp: Feature class (copy of input 2) containing three additional fields (columns) of attribute data. – Tot_Cycles – The number of harvest cycles each farm completed over the course of the simulation (duration in years) – Hrvwght_kg – Total processed weight (in kg, Eqn. 2,) for each farm summed over the time period modeled – NPV_USD_1k – The discounted net revenue from each harvest cycle summed over all harvest cycles (in thousands of $). This value will be a “0” if you did not run the valuation analysis. • Output\HarvestResults_[date and time].html: An HTML document containing tables that summarize the inputs and outputs of the model. – Farm Operations – a summary of the user-provided input data including: Farm ID Number, Weight of fish at start, Weight of fish at harvest, Number of fish in farm, start day for growing and Length of fallowing period – Farm Harvesting – a summary table of each harvest cycle for each farm including: Farm ID Number, Cycle Number, Days Since Outplanting Date, Harvested Weight, Net Revenue, Net Present Value, Outplant Day, Year – Farm Result Totals – a summary table of model outputs for each farm including: Farm ID Number, Net Present Value, Number of completed harvest cycles, Total volume harvested – Uncertainty Analysis Results – this section will be included only if uncertainty analysis was performed. It includes two parts:
* Numerical Results – a table summarizing mean and standard deviation for model outputs such as harvested weight, net present value, and number of completed harvest cycles.
* Histograms – a series of histograms to help visualize relative probabilities of different outcomes.
8.7.2 Parameter log
• Each time the model is run a text file will appear in the workspace folder. The file will list the parameter values for that run and be named according to the date and time.
8.7. Interpreting results 142 InVEST User Guide, Release 2.5.6
Figure 8.1: First few rows of a sample Farm Operations table in HTML output
Figure 8.2: First few rows of a sample Farm Harvesting table in HTML output
Figure 8.3: First few rows of a sample Farm Result Totals table in HTML output
8.7. Interpreting results 143 InVEST User Guide, Release 2.5.6
Figure 8.4: Sample histogram in the uncertainty analysis section of HTML output
8.7. Interpreting results 144 InVEST User Guide, Release 2.5.6
8.8 References
Delgado, C., N. Wada, M. Rosegrant, S. Meijer, and M. Ahmed. 2003. Outlook for Fish to 2020: Meeting Global Demand. Washington, DC: Int. Food Policy Res. Inst. FAO. 2004. Fishstat Plus. Universal software for fishery statistical series. Capture production 1950 - 2004. FAO Fish. Aqua. Dept., Fish. Inf., Data, Stat. Dep. Goldburg R., and R. Naylor. 2004. Future seascapes, fishing, and fish farming. Front. Ecol. 3:21–28. Halwart, M., D. Soto, and J.R. Arthur, J.R. (eds.) 2007. Cage aquaculture – Regional reviews and global overview. FAO Fisheries Technical Paper. No. 498. Rome, FAO. 241 pp. Naylor, R., and M. Burke. 2005. Aquaculture and Ocean Resources: Raising Tigers of the Sea. Ann. Rev. Envtl. Res. 30:185-218. Soto, D., J. Aguilar-Manjarrez, and N. Hishamunda (eds). 2008. Building an ecosystem approach to aquaculture. FAO/Universitat de les Illes Balears Expert Workshop. 7–11 May 2007, Palma de Mallorca, Spain. FAO Fisheries and Aquaculture Proceedings. No. 14. Rome, FAO. 221p. Stigebrandt, A., 1999. Turnover of energy and matter by fish—a general model with application to salmon. Fisken and Havet No. 5, Institute of Marine Research, Norway. 26 pp.
8.8. References 145 CHAPTER NINE
AESTHETIC QUALITY
9.1 Summary
The natural and scenic views of marine and coastal seascapes can contribute to the well-being of local communities in a number of ways. Scenic amenities play an important role in augmenting local economies by attracting visitors who support local businesses. The value of local property partially depends on attributes of its location and scenic views often increase local property values (Sanders and Polasky 2009, Bourassa et al. 2004, Benson et al. 2004). Local communities and their residents often become strongly attached to views and show fervent opposition to new development that has the potential to threaten the integrity of existing views and diminish the benefits drawn from those views (Ladenburg and Dubgaard 2009, Haggett 2011). The InVEST aesthetic views model allows users to determine the locations from which new nearshore or offshore features can be seen. It generates viewshed maps that can be used to identify the visual footprint of new offshore development. Inputs to the viewshed model include: topography and bathymetry, locations of offshore facilities of interest, and the locations of viewers (e.g. population centers or areas of interest such as parks or trails). The model does not quantify economic impacts of altering the viewshed, but it can be adapted to compute viewshed metrics for use in a more detailed valuation study. A key limitation of the model is that it does not currently account for the ways in which vegetation or land-based infrastructure may constrain land areas that are visually affected by offshore development. This is a “Tier 0” model.
9.2 Introduction
Coastal ecosystems are increasingly dominated by human activities. This rise in human activities can compromise the unique scenic qualities associated with coastal and marine areas. The coastline and ‘seascape’ is an important economic asset that attracts visitors for tourism and recreation and contributes to the general quality of life for people living near the coast. Near and offshore development projects often raise considerable concern within the local com- munities that value the natural seascape for its inherent beauty. Visual impacts are external effects that unless measured and accounted for, do not factor into the calculus of weighing the costs and benefits of new coastal development. Ap- plications using viewshed analysis range from the siting of aquaculture facilities to minimize spatial competition with tourism activities (Perez 2003) to seascape and shoreline visibility assessment of offshore wind projects (Environmen- tal Design and Research 2006). Because scenic beauty is an attribute generally considered to be important to people living near the coast and for those who visit coastal areas to enjoy the ocean and the marine environment, coastal plan- ners can incorporate measures of visual amenities and/or disamenities into broader policy deliberations and planning exercises. Because most applications of viewshed analysis involve examining the negative impacts of new facilities, language within the InVEST aesthetic quality model assumes the objects viewed have a negative impact on views. However, positive interpretation of viewing these objects can be included with interpretation of model results. The InVEST aesthetic quality model provides users with a simple way to provide information about potential tradeoffs between nearshore and offshore development proposals and the visual impacts of those projects. The viewshed maps produced by the model can be used to identify coastal areas that are most likely to be directly affected by additions to
146 InVEST User Guide, Release 2.5.6 the seascape. They can serve as valuable input into broader analyses that consider a range of services provided by the marine environment. Although this model does not compute the costs associated with offshore visual impacts, these costs are likely to de- crease as the location of facilities moves further offshore, while the costs of installing and operating offshore facilities generally increase with distance from the shoreline. The few valuation studies that explore the economic magnitude of visual disamenities resulting from offshore development projects show a complex picture. One recent study found that individuals living along the coast have external costs ranging from $27 to $80 resulting from the visual disamenity of an offshore wind project (Krueger et al. 2010). In contrast, Firestone et al. (2009) found that public acceptance for offshore renewable energy projects is growing and may be less contentious than previously anticipated.
9.3 The model
The aesthetic quality model provides information about the visibility of offshore objects from the surrounding land- scape or seascape. Offshore and nearshore development projects, such as renewable wave energy facilities or aquacul- ture facilities, have the potential to impact the visual amenities that are an important feature of many coastal areas. The results of viewshed analysis will be useful for decision-makers who would like to identify areas where visual impacts may be an important factor to incorporate into planning. The model requires users to provide a DEM and a point shapefile that identifies the locations of sites that contribute to visual impacts. The viewshed analysis is then computed over a user-defined area of interest (AOI) using the ArcGIS viewshed tool. The model will create as many as four outputs that can be used to assess the visible impact of any type of facility added to the marine environment. The first output, “vshed”, is a visual quality raster that records the number of sites (e.g. wave energy facilities or aquaculture farms) that are visible from a given raster cell on the land or seascape. These counts are then classify using quantiles to produce “vshed_qual” with the following class breaks: 1. Unaffected, 2. Low Visual Impact/High Visual Quality, 3. Moderate Visual Impact/Medium Visual Quality, 4. High Visual Impact/Low Visual Quality, 5. Very High Visual Impact/Poor Visual Quality. The third output computes the resident population that falls within the viewshed of any facility. The model uses the Global Rural-Urban Mapping Project (GRUMP) gridded population of the world data (CIESIN 2004) to compute the number of residents who are unaffected by the facility (or facilities) and the number of residents who live in areas that fall within the viewshed of at least one facility. The population counts are tabulated in the “populationStats.html” file found in the output folder. Users should note that this globally available population data does not account for seasonal or daily users in an area. Alternatively, you can provide your own population raster data (note that it must have the WGS84 datum). The final optional output allows for the examination of the visual impacts on areas of interest where the view is of particular concern (e.g. parks, trails, marine reserves). It utilizes a user-defined set of polygons and computes the percent area within each polygon from which at least one offshore site is visible. Each polygon is then classified by the percentage of that polygon’s area that is visually impacted by offshore developments. These results can be used to identify and rank areas according to visual impacts.
9.3.1 How it works
The InVEST aesthetic quality model is a set of wrap-around functions that employs ArcGIS’s viewshed tool. ArcGIS’s viewshed tool implements line of sight computations; the algorithm used by the tool is proprietary to ESRI and there is little documentation of the algorithm details. Users who are interested in further details should consult the ArcGIS online documentation.
9.3. The model 147 InVEST User Guide, Release 2.5.6
9.4 Limitations and simplifications
The global DEM included with the aesthetic quality model does not account for trees, buildings, or other structures that can obscure the view. If users have a raster layer that represents the locations of trees, buildings, or other obstructions (and their heights) this information can be incorporated into the DEM to create a more realistic surface to obscure or allow views. The model does account for the curvature of the earth in limiting the line of sight but it does not limit the distance at which objects of varying size and quality may be visible to the human eye in the default settings. As long as there is a straight-line vector that can be computed from a particular DEM grid cells to any offshore point, that grid cell will be counted as visible. This should be carefully considered when interpreting viewshed impact maps from facilities located far offshore when default settings are used. However, users can provide an outer radius that limits the search distance when identifying areas visible from each offshore development site.
9.5 Data needs
The model uses an interface to input all required and optional model data. Here we outline the options presented to the user via the interface and the maps and data tables used by the model. See the FAQ for detailed information on data sources and pre-processing.
9.5.1 Required inputs
First we describe required inputs. The required inputs are the minimum data needed to run this model. The minimum input data allows the model to run without conducting polygon overlap analysis. 1. Workspace (required). Users are required to specify a workspace folder path. It is recommend that the user create a new folder for each run of the model. For example, by creating a folder called “runBC” within the “Aes- theticQuality” folder, the model will create “intermediate” and “output” folders within this “runBC” workspace. The “intermediate” folder will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “output” folder. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\AestheticQuality\runBC
2. Area of Interest (AOI) (required). An AOI instructs the model where to clip the input data and the extent of analysis. Users will create a polygon feature layer that defines their area of interest. The AOI must intersect the Digital Elevation Model (DEM). Additionally, the datum of this input must be WGS84. At the start, the model will check the AOI’s datum, that it is a polygon feature and if it overlaps with the DEM input. If not, it will stop and provide feedback. Names: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\AestheticQuality\AOI_WCVI.shp
3. Point Features Impacting Aesthetic Quality (required). The user must specify a point feature layer that indicates locations of objects that contribute to negative aesthetic quality, such as aquaculture netpens or wave energy facilities. Users wish to including polygons (e.g. clear-cuts) in their analysis must convert the polygons to a grid of evenly spaced points. In order for the viewshed analysis to run correctly, the projection of this input must be consistent with the project of the DEM (input #4). At the start, the model will check that inputs #3 and #4 have consistent projections. If not, it will stop and provide feedback. For instructions on how to create a point shapefile, see the InVEST FAQ. Names: File can be named anything, but no spaces in the name File type: point shapefile (.shp) Sample path: \InVEST\AestheticQuality\AquaWEM_points.shp
9.4. Limitations and simplifications 148 InVEST User Guide, Release 2.5.6
4. Digital Elevation Model (DEM) (required). A global raster layer is required to conduct viewshed analysis. Elevation data allows the model to determine areas within the AOI’s land-seascape where features from input #3 are visible. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: standard GIS raster file (e.g., ESRI GRID or IMG), with elevation values Sample data set: \InVEST\AestheticQuality\Base_Data\Marine\DEMs\claybark_dem
5. Refractivity Coefficient (required). The earth curvature correction option corrects for the curvature of the earth and refraction of visible light in air. Changes in air density curve the light downward causing an observer to see further and the earth to appear less curved. While the magnitude of this effect varies with atmospheric conditions, a standard rule of thumb is that refraction of visible light reduces the apparent curvature of the earth by one-seventh. By default, this model corrects for the curvature of the earth and sets the refractivity coefficient to 0.13. Names: A string of numeric text with a value between 0 and 1 File type: text string (direct input to the ArcGIS interface) Sample (default): 0.13
9.5.2 Optional inputs
The next series of inputs are optional, but may be required depending on other decision inputs. 6. Cell Size (meters) (optional). This determines the spatial resolution at which the model runs and at which the results are summarized. For example, if you want to run the model and see results at a 100m x 100m grid cell scale then enter “100.” You can only define a resolution that is equal to or coarser than the model’s native resolution as established by the current DEM (input # 4). If you want to run the model and produce output at the current DEM’s resolution (the model’s native resolution) you can leave this input field blank. The coarser the scale (and larger the number), the faster the model runs. Names: A numeric text string (positive integer) File type: text string (direct input to the ArcGIS interface) Sample (default): 500
7. Global Population Raster (required). A global raster layer is required to determine population within the AOI’s land-seascape where features from input #3 are visible and not visible. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: standard GIS raster file (ESRI GRID) with population values Sample data set (default): \InVEST\Base_Data\Marine\Population\global_pop
8. Polygon Features for Overlap Analysis (optional). The user has the option of providing a polygon feature layer where they would like to determine the impact of points (input #3) on visual quality. This input must be a polygon and projected in meters. The model will use this layer to determine what percent of the total area of each feature can see at least one of the points from input #3. Names: File can be named anything, but no spaces in the name File type: polygon shapefile (.shp) Sample path: \InVEST\AestheticQuality\BC_parks.shp
9.6 Running the model
Note: The word ‘path‘ means to navigate or drill down into a folder structure using the Open Folder dialog window that is used to select GIS layers or Excel worksheets for model input data or parameters.
9.6. Running the model 149 InVEST User Guide, Release 2.5.6
9.6.1 Exploring the workspace and input folders
These folders will hold all input, intermediate and output data for the model. As with all folders for ArcGIS, these folder names must not contain any spaces or symbols. See the sample data for an example.
Exploring a project workspace and input data folder
The \InVEST\AestheticQuality folder holds the main working folder for the model and all other associated folders. Within the AestheticQuality folder there will be a subfolder named ‘Input‘. This folder holds most of the GIS and tabular data needed to setup and run the model. The following image shows the sample folder structure and accompanying GIS data. We recommend using this folder structure as a guide to organize your workspaces and data. Refer to the following screenshots below for examples of folder structure and data organization.
9.6.2 Creating a run of the model
The following example of setting up the Aesthetic Quality model uses the sample data and folder structure supplied with the InVEST installation package (see the Data needs section for a more complete description of the data). These instructions only provide a guideline on how to specify to ArcGIS the various types of data needed and does not represent any site-specific model parameters. Users might choose different input parameters and/or have location- specific data to use in place of the sample data. 1. Click the plus symbol next to the InVEST toolbox. 2. Expand the Marine toolset and click on the Aesthetic Quality script to open the model. 3. Specify the Workspace. Open the InVEST workspace. If you created your own workspace folder (Step 1), then select it here. Select the AestheticQuality folder and click to set the main model workspace. This is the folder in which you will find the intermediate and final outputs when the model is run. 4. Specify the Area of Interest (AOI). The AOI is the geographic area over which the model will be run. This example refers to the AOI_WCVI.shp shapefile supplied in the sample data. You can create an AOI shapefile by following the Creating an AOI instructions in the FAQ. Open the \InVEST\AestheticQuality\Input data folder.
If you created your own Input folder in step 1b, then select it here. Select the AOI shapefile and click to make the selection.
9.6. Running the model 150 InVEST User Guide, Release 2.5.6
9.6. Running the model 151 InVEST User Guide, Release 2.5.6
5. Specify the Cell Size. This option determines the cell size for the output viewshed raster. The default is “500”, meaning the model will run at the 500m resolution utilizing the input DEM. You can type directly into the text box to specify a different value. 6. Specify the Point Features Impacting Aesthetic Quality. This vector dataset represents points that have undesir- able effects on aesthetic viewing quality. Open the Input data folder \InVEST\AestheticQuality\Input and click the AquaWEM_points.shp shapefile. 7. Specify the Digital Elevation Model. The digital elevation model provides the base upon In- VEST\Base_Data\Marine\DEMs folder, select the claybark_dem raster and click . 8. Specify the Refractivity Coefficient. The model requires a refractivity coefficient. The default value is value 0.13. You can type directly into the text box to specify a different value. 9. Specify Global Population Raster. This dataset represents raster cells of population and is required for the viewshed analysis. Open the \InVEST\Base_Data\Marine\Population folder and click the global_pop raster. 10. Specify Polygon Features for Overlap Analysis (Optional). This vector dataset represents polygon areas to be considered for the viewshed analysis. Open the \InVEST\AestheticQuality\Input data folder and add the BC_parks.shp shapefile. 11. At this point the model dialog box is completed for a complete run of the Aesthetic Quality model. Click to start the model. The model will begin to run and a show a progress window with progress information about each step in the analysis. Once the model finishes, the progress window will show all the completed steps and the amount of time necessary for the model run.
9.6. Running the model 152 InVEST User Guide, Release 2.5.6
9.6. Running the model 153 InVEST User Guide, Release 2.5.6
9.6.3 Multiple runs of the model
The model setup is the same as for a single run, but the user needs to specify a new workspace for each new run. Make sure each new workspace exists under the main model workspace folder (i.e. AestheticQuality folder in the example above). As long as all data are contained within the main Input data folder you can use the same Input folder for multiple runs. For example, using the sample data, if you wanted to create two runs of the Aesthetic Quality model based on two different visual polygon shapefiles (BC_parks.shp and BC_protectedAreas.shp), you could use the Input data folder under main AestheticQuality folder and create two new workspace folders, BC_parks and BC_protectedAreas. See below for an example of the folder setup.
9.6.4 Viewing output from the model
Upon successful completion of the model, you will see new folders in your Workspace called “intermediate” and “Out- put”. The Output folder, in particular, may contain several types of spatial data, which are described the Interpreting results section.
You can view the output spatial data in ArcMap using the Add Data button .
You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting “Properties”, and then “Symbology”. There are many options here to change the way the data appear in the map. You can also view the attribute data of output files by right clicking on a layer and selecting “Open Attribute Table”.
9.7 Interpreting results
9.7.1 Model outputs
The following is a short description of each of the outputs from the aesthetic views model. Each of these output files is saved in the “Output” folder that is saved within the user-specified workspace directory:
9.7. Interpreting results 154 InVEST User Guide, Release 2.5.6
Output folder
• Output\vshed_qual – This raster layer contains a field that classifies based on quartiles the visual quality within the AOI. The visual quality classes include: unaffected (no visual impact), high (low visual impact), medium (moderate visual impact), low (high visual impact), and very low (very high visual impact). – Additionally, the range of sites visible for each visual quality class is specified in this output’s attribute table. – This layer can be symbolized by importing the symbology from the file “\AestheticQual- ity\Input\vshed_qual.lyr” • Output\vshed – This raster layer is the original output after the viewshed tool is run. It contains values ranging from 0 to the total number of points visible from each cell on the land or seascape. For example, all cells with a value of “4” would indicate that at that location four points are visible. – In order to compare scenario runs, use this layer rather than vshed_qual. By calculating the difference between “vshed” outputs from multiple runs, a user can assess changes in visual quality across scenarios. • Output\vp_overlap.shp – This polygon feature layer contains a field called “AreaVShed” which expresses the percentage of area within each polygon where at least one point contributing to negative aesthetic quality is visible as com- pared to the total area of that polygon. – This layer can easily be symbolized by importing the symbology from the file “\AestheticQual- ity\Input\vp_overlap.lyr” • Output\populationStats_[date and time].html – This html file includes a table and indicates the approximate number of people within the AOI that are 1) unaffected (no sites contributing to negative aesthetic quality are visible) and 2) affected (one or more sites visible). • Parameters_[yr-mon-day-min-sec].txt – Each time the model is run a text file will appear in the workspace folder. The file will list the parameter values for that run and be named according to the date and time.
Intermediate folder
• intermediate\dem_vs – This raster layer is the modified DEM within the user-specified extent. The portions of the DEM that are below sea-level are converted to a value of “0” since all viewing on the ocean will be at the surface.
9.8 Case example illustrating results
The following example illustrates the aesthetic views model. In this example, we examine the visual footprint resulting from potential wave energy facilities and aquaculture farms. The following figures and maps are for example only, and are not necessarily an accurate depiction of WCVI. In the first figure, we show the locations of the sites of potential wave energy facilities and aquaculture farms. In this example, there are four offshore wave energy facilities and ten aquaculture facilities. We then run the aesthetic views model to determine the visual footprint of these potential facilities. To run the model, we first create an area
9.8. Case example illustrating results 155 InVEST User Guide, Release 2.5.6
of interest polygon that encompasses all of the site locations and the portion of the sea and landscape that we are interested in evaluating. We then apply an upper bound of 8 km on the search radius. This limits the search distance to 8 km when identifying areas that are visible from each observation point. This upper bound is applied by adding the field RADIUS2 to the shapefile specifying the point features contributing to negative aesthetic quality. To limit the search to 8 km, each point is assigned a value of -8000 as shown in the following figure.
After completing the steps outlined in the “Running the model” section, we obtain the following map that classifies the visual impacts of these sites.
9.8.1 Classification of visual quality
The resulting map shows the footprint of visual quality from offshore wave energy sites and the aquaculture facilities. The cells highlighted in red are the areas with the highest visual impact; the cells highlighted in green have the lowest visual impact. The grey cells have no visual impact. It is clear from the visual quality map that most offshore areas experience low visual impacts from the wave energy facilities, whereas areas surrounding the clustered aquaculture facilities experience the highest visual impacts. Please be aware that the quality of the viewshed model results depends
9.8. Case example illustrating results 156 InVEST User Guide, Release 2.5.6 on the quality of the DEM used in the analysis. Fine resolution DEMs that account for trees, buildings, and other obstructions will give the most realistic results.
9.8.2 Resident population impacted by visual disamenities
In addition to producing a map of the visual footprint of objects located offshore, the aesthetic quality model also provides a count of the resident population that falls within this visual footprint. The viewshed model uses the Gridded Rural-Urban Population Model Project (GRUMP) dataset to extract the population counts within grid cells that are visible from any of the offshore sites. These counts are then tabulated and documented in the “PopulationStats.html” file found in the output folder. For this example, the number of residents unaffected by the offshore sites is 8554
and the population count that falls within grid cells that can see at least one offshore site is 3735. Users again should be reminded that the GRUMP dataset is based on site-specific census data and may not accurately reflect the actual
9.8. Case example illustrating results 157 InVEST User Guide, Release 2.5.6 population that uses a particular area. This is particularly true for areas important for tourism and other seasonal activities that census data will not account for.
9.8.3 Viewshed overlap with protected areas
The final optional output of the aesthetic quality tool uses a set of user-specified polygons and computes the percent area within each polygon from which at least one offshore site is visible. To illustrate these results, we use a set of polygons that represent protected areas in the same study area explored above.
The protected areas are shown in the above figure as green polygons and the points represent the location of the offshore wave energy facilities and aquaculture sites. For each protected area in the user-specified area of interest, the model then computes the percentage of each protected area that falls within the viewshed of the wave energy and aquaculture sites. The figure below shows the results for a selection of the protected areas included in the example.
From this example, we see that for most of the protected areas, 1-25% of their total area falls within the viewshed footprint of the wave energy and aquaculture sites. For one of the smaller protected areas, 51-75% of its area falls within the viewshed footprint. These results are not spatially explicit at a fine scale because they do not indicate the exact locations from which one could see the facilities. However, these locations can be identified from the previous aesthetic quality results.
9.8. Case example illustrating results 158 InVEST User Guide, Release 2.5.6
9.9 References
Benson E., Hansen, J., Schwartz, A., and Smersh, G., 1998. Pricing residential amenities: the value of a view. Journal of Real Estate Research, 16: 55-73. Bourassa, S., Hoesli, M. and Sun, J. 2004. What’s in a view? Environment and Planning A. 36(8): 1427-1450. Center for International Earth Science Information Network (CIESIN), Columbia University; International Food Pol- icy Research Institute (IFPRI); The World Bank; and Centro Internacional de Agricultura Tropical (CIAT). 2004. Global Rural-Urban Mapping Project (GRUMP), Alpha Version: Population Grids. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University. Available at http://sedac.ciesin.columbia.edu/gpw. (downloaded on 1/6/2011). Environmental Design and Research, P.C. 2006. Seascape and shoreline visibility assessment. Cape Wind Energy Project. Cape Cod, Martha’s Vineyard, and Nantucket, Massachusetts. Prepared for Cape Wind Associates, L.L.C. Boston, Mass. Syracuse, N.Y. July 2006. Firestone, J., Kempton, W. & Krueger, A., 2009. Public acceptance of offshore wind power projects in the USA. Wind Energy, 12(2):183-202. Haggett, C. 2011. Understanding public responses to offshore wind power. Energy Policy. 39: 503-510. Krueger, A., Parson, G., and Firestone, J., 2010. Valuing the visual disamenity of offshore wind power at varying distances from the shore: An application of on the Delaware shoreline. Working paper. Available at: http://works.bepress.com/george_parsons/doctype.html. Ladenburg, J. & Dubgaard, A., 2009. Preferences of coastal zone user groups regarding the siting of offshore wind farms. Ocean & Coastal Management, 52(5): 233-242. Perez, O.M., Telfer, T.C. & Ross, L.G., 2005. Geographical information systems-based models for offshore floating marine fish cage aquaculture site selection in Tenerife, Canary Islands. Aquaculture Research, 36(10):946-961. Sander, H.A. & Polasky, S., 2009. The value of views and open space: Estimates from a hedonic pricing model for Ramsey County, Minnesota, USA. Land Use Policy, 26(3):837-845.
9.9. References 159 CHAPTER TEN
OVERLAP ANALYSIS MODEL
10.1 Summary
Mapping current uses and summarizing the relative importance of various regions for particular activities is an impor- tant first step in marine spatial planning. The InVEST Overlap Analysis Model was designed to produce maps that can be used to identify marine and coastal areas that are most important for human use. Initial development of this model was as two separate models for recreation and fisheries. However, since the underlying approach was fundamentally similar, we combined them into one model that can be used to map not only recreation and fisheries, but also other activities. While this model was envisioned for use in marine areas where space is shared commonly, it is also appli- cable to locations on land where overlapping uses occur. Inputs include information about where human use activities occur (required), weights that reflect the relative importance of different human uses (optional) and information on spatial variability within uses (optional). Because it simply maps current uses and does not model behavior, this model is not well-suited to the evaluation of how human uses may change in response to changes in the coastal and marine environment. However, it can be used to model scenarios that reflect changes in the areas used by different activities or changes in attributes such as total landings or number of trips that are used to weight activities.
10.2 Introduction
Understanding where and how humans use coastal and marine environments is an essential component of marine resource planning and management. Marine and coastal ecosystems are essential places for a variety of activities in- cluding fishing (commercial, recreational, subsistence and ceremonial) and recreation (e.g. boating, kayaking, diving, whale-watching). When siting new activities and infrastructure or zoning areas for particular uses, a key step is the identification and visualization of the variety of human uses that occur in the region and the places in which they overlap (e.g., GBRMPA 2003, CDFG 2008, Beck et al. 2009, CRMC 2010). This allows for the identification of hotspots of human use and highlights regions where the compatibility of various activities should be investigated. The InVEST Overlap Analysis model provides users with a simple framework for mapping and identifying important areas for human use in the marine environment. The model also allows users to include information about a variety of uses of the coastal and marine environment (e.g., commercial fishery logbooks or landings reports data, participation numbers for recreational activities) that can be used to weigh the relative importance of different uses and locations. The model is simple to use, quick to run, and can be applied in any region of the world where there is spatially-explicit information on human uses. The model does not value environmental services or estimate the economic value of human uses, but the outputs can be used to identify areas and different user groups that may be affected by policy change. The model produces a map of hotspots for human activities (e.g., fishing activity/fishing grounds) across as many human uses as the user chooses to include. Throughout this chapter we will give examples for both recreation and fisheries. Using the tool across various categories of human uses may make sense in some instances, but devising schemes for weights will likely be difficult. Outputs can be used to help decision-makers weigh potential conflicts between sectors of spatially-explicit management options that may involve new activities or infrastructure.
160 InVEST User Guide, Release 2.5.6
The Overlap Analysis Model complements more involved InVEST fisheries and recreation models that are in devel- opment. The InVEST Marine Fish Aquaculture model is appropriate for use with single species or groups of species and is used to estimate the quantity and value of fish harvested by commercial fisheries. Additionally, a recreational submodel can be used to predict the amount of recreational fishing effort required to catch the quantity of fish output from the InVEST Marine Fish Aquaculture model. Future more advanced fisheries models will include functionality to incorporate impacts of biogenic habitat on the survival and fecundity of different life-stages of target species, and the ability to wrap around outputs from more complex food-web models (e.g., Ecopath with Ecosim and Atlantis).
10.3 The model
The InVEST Overlap Analysis model was designed to identify marine and coastal areas most important for human use. The model combines the different input layers of human use and computes an “Importance Score” for each grid cell or management area. If users only know where activities occur but do not have additional information to weight the relative importance of different activities, the default model computes an “Importance Score” by summing the number of activities that occur in any particular cell or zone. Although not required to run the model, users can input qualitative (e.g., indices, scores) or quantitative (e.g., catches, effort levels, revenues, profits) information to weight the importance of different locations for an individual activity and to weight activities compared to one another. The model also allows users to down-weight areas or zones used for different activities as a function of their distance from important land-based hubs such as ports, marinas, or public access points. Model outputs are mapped in the coastal region of interest over the specified seascape or management zones. The default model map output is a shapefile showing the frequency of occurrence of activities across the area of interest. If additional weighting information is included, the model also produces a shapefile showing the gradation of importance across cells or zones. The resulting maps can then be used to evaluate the relative importance of different areas in the seascape for the set of human activities included in the analysis. See Appendix A for suggestions for data sources.
10.3.1 How it works
Calculating frequencies (model default)
Users input maps of the locations of multiple human activities. Data is input in a vector format as polygons or points; vector data are rasterized after they are input. In the simplest (default) model, all activities and locations are weighted equally and the model calculates an Importance Score (IS), which is a count of how many activities take place in each grid cell or management zone i: X ISi = UijIj (10.1) i,j
where Uij = usage of activity j in grid cell or management zone i. Uij is scored by the presence (Uij = 1) or absence (Uij = 0) of the activity in the cell or zone.
Including weights (optional)
Users are also given the option to apply different weights to each activity. The two ways in which users can provide these weights are as inter- or intra-activity weights: 1. Inter-activity weight: this allows users to weight the importance of activities relative to one another. Users may choose to give more weight in the analysis to certain activities (e.g., those that generate the highest profits of all fleets in the analysis, or are key employers in the region) and less to other activities. For example, if the user is examining 3 activities (1. commercial salmon fishing, 2. commercial crab fishing, and 3. commercial kelp harvest) and commercial salmon fishing is deemed to be twice as important as either commercial crab fishing
10.3. The model 161 InVEST User Guide, Release 2.5.6
or commercial kelp harvest, then the user would provide weights of (commercial salmon fishing, commercial crab fishing, commercial kelp harvest)= (2,1,1). Inter-activity weights are included in an input .csv table (see “Running the Model” section below); and/or 2. Intra-activity weight: Spatially explicit information about the relative importance of various locations (points or polygons on the map) for a particular activity can be used to weight the scores used in the model calculations. Importance can be measured several ways. For fisheries, weights might be informed by the amount of fish caught or landed, profits earned, safety or accessibility of the fishing ground, or the cultural value of the area. For recreation, they might be determined by the number of visitors or trips to different areas. For example, if the user is examining three commercial harvesting activities and has catch data for each polygon representing those activities, these intra-activity weights can be included by adding a column to the shapefile attribute table of each input activity layer. The name of this column should have no spaces, and this column name will need to be given as an input so that the model knows where these weights are stored. If intra- or inter-activity weights are included, IS is weighted by the importance of the cell (or zone) relative to other cells (or zones) with that activity occurring, and/or the importance of the activity relative to other activities included in the analysis. Please see Appendix A for guidance on preparing and including information on intra- and inter-activity weights using qualitative (i.e., scores of ‘more’ or ‘less’ fishing in a cell, visitation or trip numbers for recreational activities) or quantitative (i.e., commercial fishing catch, effort level, revenues, profits) data. Functionally, IS of pixel or management zone i is:
1 X IS = U I i n ij j (10.2) i,j
where: n = number of human use activities included in the analysis.
Uij = usage or intra-activity weight (optional) of activity j in pixel or management zone i. If the user does not include intra-activity weights (i.e., model default), Uij represents usage and is scored by presence (Uij = 1) or absence (Uij = 0) of the activity in the cell or zone. When intra-activity weights are included, Uij reflects the weights as Uij = Xij / Xmaxj, where Xij is the intra-activity weight of activity j in pixel or management zone i and Xmaxj is the maximum intra-activity weight for all cells or zones where the activity occurs.
Ij = inter-activity weight (optional) of activity j relative to other activities included in the analysis. If the user treats all activities as equally important (model default), Ij is ignored (i.e., Ij = 1). When inter-activity weights are included, Ij reflects the weights as Ij = Yj / Ymax, where Yj is the inter-activity weight of activity j and Y max is the maximum inter-activity weight for all activities.
10.4 Limitations and simplifications
This model is a very simple framework that provides little insight into how human activities might change under different scenarios of change in the coastal and marine environment. Such insights are best gleaned from models that include descriptors of human behavior. However, scenarios that add or remove activities or change weights of various activities and/or locations can be used to explore change.
Warning: the model is very sensitive to inter- and intra-activity weights. Therefore, the assumptions you make when including these optional inter- and/or intra-activity weights will strongly affect model outputs. If you are unsure of how to appropriately include inter- or intra-activity weights, we encourage you to conduct several model runs to see how different weighting schemes affect model outputs.
10.4. Limitations and simplifications 162 InVEST User Guide, Release 2.5.6
10.5 Data needs
The model uses an interface to input all required and optional model data. There are two options: the standard overlap analysis model that computes use intensity for each raster pixel, and an option to compute intesity by larger management zones. Each is a standalone model in InVEST, however the inputs required have the same descriptions and requirements so they are not reft below. Here we outline the options presented to the user via the interface, and the maps and data tables used by the model. First we describe required inputs, followed by a description of optional inputs.
10.5.1 Required inputs
The required inputs are the minimum data needed to run this model. The minimum input data allows the model to run without importance weights or distance decay, both of which are optional parameters. 1. Workspace Location (required). Users are required to specify a workspace folder path. We recommend that you create a new folder for each run of the model. For example, by creating a folder called “runBC” within the “OverlapAnalysis\Recreation” folder, the model will create “intermediate” and “output” folders within this “runBC” workspace. The “intermediate” folder will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “output” folder. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\OverlapAnalysis\BCrun
2. Analysis Zones Layer (required). A polygon shapefile that defines the area of interest for the standard analysis. The AOI must be projected with linear units equal to meters. For the management zones model, a similar shapefile is needed except the AOI should be divided into appropriate management zones. Name: File can be named anything, but no spaces in the name File type: Polygon shapefile (.shp) Sample path: \InVEST\OverlapAnalysis\Input\AOI_WVCI.shp
3. Analysis Cell Size (required). This determines the spatial resolution at which the model runs and at which the results are summarized. For example, if you want to run the model and see results at a 100m x 100m pixel size then enter “100.” Name: A numeric text string (positive integer) File type: text string (direct input) Sample (default): 1000
4. Overlap Analysis Data Directory (required). Users are required to specify the path on their system to a folder containing only the input data for the Overlap Analysis model. Input data can be point, line or polygon data layers indicating where the human use activity takes place (e.g., whale watching, diving, or kayaking in a marine setting). For instructions on how to create a polygon or raster shapefile, see the FAQ. Please note that optional intra-activity importance information, described below for optional input #1, can be associated with each layer. Name: Path to an activity data folder. Avoid spaces. Sample path: \InVEST\OverlapAnalysis\Input\RecreationLayers_RIS\
Note: All data in this folder must be shapefiles and projected in meters.
10.5.2 Optional inputs
The next series of inputs are optional for added model functionality.
10.5. Data needs 163 InVEST User Guide, Release 2.5.6
1. Intra-Activity Attribute Name (optional). The user has the option of providing information on the importance of locations (i.e., polygons or points) within a layer of human use data (e.g., one fishing ground may be much more valuable than another; certain kayaking routes may be more popular than others). These intra-activity importance scores can be qualitative or quantitative (see Appendix for further description of data inputs) and must be listed in a new column of the attribute tables for each layer included in the Overlap Analysis (see intra-activity weighting in The model section). The name given to the column that contains the intra-activity importance scores must be the same for all layers contained within the directory specified by input #4. The model uses this information to weight the importance of areas found within each input layer. Names: Text string containing letters and/or numbers (must start with a letter). Field name must correspond to an existing column name in each layer’s attribute table Sample: RIS
2. Inter-Activity Weight Table (optional). The model also allows users to provide information on the relative importance of uses. This .csv file lists the activities and gives them a numerical relative importance weighting. The default files demonstrate the required structure; it is recommended that these files not be overwritten. In the .csv table, it is important that the name of each use exactly corresponds to the given name of the shapefile that represents that use. Names: File can be named anything, but no spaces in the name File type: Comma-separated values file (.csv) Sample path: \InVEST\OverlapAnalysis\Input\Recreation_Inputs.csv
3. Points Layer of Human Use Hubs (optional). The model allows users to down-weight areas or zones used for different activities as a function of the distance from important land-based hubs such as ports, marinas, or public access points. This input GIS layer must be a point shapefile and projected in meters. Names: File can be named anything, but no spaces in the name File type: Point shapefile (.shp) Sample path: \InVEST\OverlapAnalysis\Input\PopulatedPlaces_WCVI.shp
4. Distance Decay Rate (optional). If a GIS layer is specified for optional input #3, the model will use a decay rate of β =0.025 by default. If this input is not specified, no distance decay occurs and this rate is ignored. See Figure 1 for how changing this parameter changes the decay rate. With a decay rate of 0.025, an importance score of 1 would decrease to ~0.8 at a distance of approximately 10 km from the nearest hub. User judgment should be exercised when using this option. The following scenario illustrates one example of how users might use the distance decay function. Suppose you know that the intensity of human activities is greatest in areas relatively close to the ports, marinas, and other public access points, but you do not have the data necessary to construct spatially-explicit weighting factors to reflect this knowledge. In the absence of these data, the distance decay function could be used to reflect this intensity / distance tradeoff. You can choose a decay rate that reflects your best judgment on how the importance (e.g., intensity) of activities declines with distance from important population centers, marinas, or access points. For example, if most recreational fishing grounds are located within 10 km from the central marina, you could choose a decay parameter of β =0.01 to reflect a gradual threshold in the decline of importance of more distant sites, or β =0.5 to reflect a sharper threshold. Names: A string of numeric text with a value between 0 and 1 File type: Text string (direct input to the ArcGIS interface) Sample (default): 0.025
Exponential decay functions used to downweight importance of activities based on distance from land-based access point
10.5.3 Multiple runs of the model
The tool setup is the same as for a single run, but you must specify a new workspace for each new run. Make sure each new workspace exists under the main workspace folder (i.e. OverlapAnalysis folder in the example above). As
10.5. Data needs 164 InVEST User Guide, Release 2.5.6
10.5. Data needs 165 InVEST User Guide, Release 2.5.6 long as all data are contained within the main Input data folder you can use the same Input folder for multiple runs. For example, using the sample data, if you wanted to create two runs of the Overlap Analysis model based on two different weighting systems for fishing fleets, you could use the Input data folder under main Overlap Analysis folder and create two new workspace folders, runFisheries1 and runFisheries2.
10.5.4 Viewing output from the model
Upon successful completion of the model, you will see new folders in your Workspace called “intermediate” and “Output”. The Output folder, in particular, will contain several types of spatial data, which are described in the Interpreting Results section.
You can view the output spatial data in ArcMap using the Add Data button.
You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting “Properties”, and then “Symbology”. There are many options here to change the way the data appear in the map. You can also view the attribute data of output files by right clicking on a layer and selecting “Open Attribute Table”.
10.6 Interpreting results
10.6.1 Model outputs
The following is a short description of each of the outputs from the Overlap Analysis model. Each of these output files is saved in the “Output” folder that is saved within the user-specified workspace directory:
Output folder
• Output\hu_freq
10.6. Interpreting results 166 InVEST User Guide, Release 2.5.6
– This raster layer depicts the frequency of activities for each cell or management zone for the study area. Each layer input is only counted once regardless of the number of features within that layer overlapping a cell. Therefore, if three layers are specified in the input directory, then the max value of this output is 3. – This is the default model output that will be generated for each run of the model. • Output\hu_impscore – This raster layer depicts Importance Scores for each cell or management zone for the study area. – This output is only generated if the user includes intra-activity weights defined by optional input #1: “Importance Score Field Name”. • overlap_analysis-log-yr-mon-day-min-sec].txt – Each time the model is run a text file will appear in the workspace folder. The file will list log information that can be used to identify detailed configurations of each of scenario simulation.
10.7 Appendix A
10.7.1 Preparing input data
Maps of fishing grounds
Users should create a layer of polygons or points to define where individual fishing fleets operate. Fleets can be defined however you deem appropriate. Often, fleets are defined by their sector (e.g., commercial, recreational, subsistence), the species or species complexes they target (e.g., prawn, salmon, groundfish), and the gear that they use (e.g., trawl, seine, longline). For example, fleets might be commercial groundfish trawl, subsistence salmon seine, or recreational tuna hook and line. For each fleet you decide to include, you must have information on where that fleet fishes. Locations can be points or polygons. You can generate these layers if existing maps of spatial distribution of fishing catch or effort are available to you. These maps are not often readily available, in which case, you can summarize catch, effort, or revenue data by management zone or statistical area. Availability of these data varies regionally – most regional management councils in the U.S. collect these data and make them publicly available through data clearinghouses associated with regional management councils (e.g., Pacific Fisheries Information Network associated with Pacific Fisheries Management Council). When summary by management zone or statistical area is unavailable, information can be solicited from stakeholders through exercises where they draw polygons or points on maps. If none of these are options for you, but you have habitat information available, it is possible to draw habitat-species-gear associations and coarsely estimate where fleet activity may occur.
Recreational activity layers
Spatially explicit data on recreation activities can be collected from a variety of sources including local tourism opera- tors, government agencies, and guide books. In most areas, there is no clearinghouse for this type and users will likely need to combine data from a variety of sources.
Importance data (optional)
Intra-fleet weights
Quantitative or qualitative or data on which locations in the coastal and marine environment are most or least important for a human use (i.e., intra-activity weights) can be easily prepared and included in the Overlap Analysis model.
10.7. Appendix A 167 InVEST User Guide, Release 2.5.6
Whichever type of data is used does not need to be consistent across human use activities. For example, when spatially- explicit catch data exist for one fishing fleet, and another fishing fleet only has qualitative rankings of importance of different fishing grounds, both data sets can be used. Intra-fleet weights are entered for each polygon or point in each data layer’s attribute table. If intra-fleet weights are missing for one or more data layers in the analysis, users must include a placeholder column (i.e., values for all polygons in the layer = 1) for the model to run correctly. Quantitative data are likely to be catch, effort, profit, or revenue information for fisheries. For recreation, the number of trips or number of visitors to each site is the suggested metric to be used to weight activities. Alternatively, users may use the number of days that an area is open to particular activities or other metrics that proxy for importance or usage. Higher values should indicate polygons or points of higher importance than those with lower values. Qualitative scoring is a good option for users without quantitative input data. Low scores should indicate least impor- tant locations for the activity, high scores most important areas, and multiple areas should be allowed to have the same score (i.e., areas are given scores, not ranks). We encourage users to take care in assignment of values to locations as these values strongly influence outcomes. For example, if one fishing area polygon is given a score of 1, and another a score of 2, is the 2nd polygon twice as “important” as the first? If not, and the two polygons are more similar in their importance, the user could considering scoring more closely to one another (e.g., score of 1.75 and 2, instead of 1 and 2) or score on a larger scale (e.g., scores of 4 and 5, instead of 1 and 2). The onus is on the user to decide which range of weights to use. If you are unsure of how to appropriately include these weights, we encourage you to conduct several model runs to see how different weighting schemes affect model outputs. A common method for obtaining qualitative information on the importance of an activity is by querying stakeholders or decision-makers in the region. InVEST will soon include a mapping tool to help collect data from stakeholders. The tool will include functional- ity for entering intra-activity weights. If using the InVEST drawing tool (forthcoming) while querying stakeholders, importance scores can be input when generating layers. Once intra-activity weights are input into the model, they are scaled by the maximum value for all locations where the activity occurs. For example, if the user has identified 3 fishing grounds for a fleet, with values of 2, 4, and 5, they will be scaled by 5, to be 0.4, 0.8 and 1.0.
Inter-activity weights
The user has the option to include information on the importance of activities relative to one another so that all activities are not treated equally. This information is not spatially explicit, rather is in the form of one value for each activity. If the user chooses to include inter-activity weights, they must be included for all activities. Inter-activity weights can be qualitative (e.g., stakeholder designated) or quantitative (e.g., total catch, effort, profit, or revenue; socio-economic assessment of contributions of each fishing fleet to community stability or tax base), but the same metric should be used to weight all activities. For recreation, if the user does not have spatially explicit data on numbers of recreation trips, but does have the aggregate number of trips or participants for each activity, these numbers can be used to construct an importance ranking of each activity by using the percentage of trips / participants in each activity as inter-activity weights. For fisheries, for example, if running the model for three fishing fleets, inter-activity weights could be calculated using total revenue earned by each fleet as is done in the example presented earlier in this chapter. It would be inappropriate to determine weights by comparing one fleet’s catches to the others’ revenues. Given this caution, when determining inter-activity weights, users should choose a common quantitative (e.g., catch, revenue for fishing fleets) or qualitative (e.g., scores from stakeholder input) metric that is applicable across all activities. Similar to the intra-activity weights, inter-activity weights are not ranks (i.e., activities can have the same weights), and must be included for all data layers. Once input into the model, quantitative or qualitative values are scaled by the maximum value for all activities. The caution in the preceding, intra-activity, section about the numeric scales used for qualitatively weighting activities applies here, as weights strongly affect model outputs. To reiterate, using a hypothetical model run for recreational data, if the inter-activity weight for whale-watching is 1, and kayaking 2, is the kayaking twice as “important” as whale-watching? If the activities are actually more similar, the weights should be closer to one another (e.g., score of 1.75 and 2, instead of 1 and 2) or score on a larger scale (e.g., scores of 4 and 5, instead of 1 and 2). Users are responsible for choosing the range of weights to use, and we encourage you to conduct several model runs to see how different weighting schemes affect model outputs.
10.7. Appendix A 168 InVEST User Guide, Release 2.5.6
10.8 References
Beck, M.W, Z. Ferdana, J. Kachmar, K. K. Morrison, P. Taylor and others. 2009. Best Practices for Marine Spatial Planning. The Nature Conservancy, Arlington, VA. 32 pp. CDFG (California Department of Fish and Game). 2008. California Marine Life Protection Act. Master Plan for Marine Protected Areas. 110 pp. CRMC (Coastal Resources Management Council). 2010. Rhode Island Ocean Special Area Management Plan: Adopted by the Rhode Island Coastal Resources Management Council October 2010. 993 pp. DFO (Department of Fisheries and Oceans). 2008. Canadian Fisheries Statistics 2006. Ottawa: Fisheries and Oceans Canada. GBRMPA (Great Barrier Reef Marine Park Authority). 2003. Great Barrier Reef Marine Park Zoning Plan 2003. Australian Government. 220 pp.
10.8. References 169 CHAPTER ELEVEN
HABITAT RISK ASSESSMENT
11.1 Summary
The condition of a habitat is a key determinant of the environmental services it can provide. For example, multiple stressors including fishing, climate change, pollution and coastal development threaten the ability of coastal ecosys- tems to provide the valuable goods and services that people want and need. As human activities continue to intensify, so too does the need for quick, clear and repeatable ways of assessing the risks posed by human activities under various management plans. The InVEST habitat risk assessment (HRA) model allows users to assess the risk posed to coastal and marine habitats by human activities and the potential consequences of exposure for the delivery of environmental services and biodiversity. The InVEST HRA model is similar to the InVEST biodiversity model in that both models allow users to identify regions on a landscape or seascape where human impacts are highest. While the biodiversity model is intended to be used to assess how human activities impact biodiversity, the HRA model is better suited to screening the risk of current and future human activities to prioritize management strategies that best mitigate risk. We built and tested the HRA model in marine and coastal systems, and discuss it accordingly, but it easily can be applied to terrestrial systems, or mobile species. Risk of human activities (e.g., salmon aquaculture, coastal development, etc.) to habitats (e.g., seagrasses, kelp forests, mangroves, reefs) is a function of the exposure of each habitat to each activity and the consequences for each habitat. Exposure to stressors can arise through direct overlap in space and time or through indirect effects (i.e. finfish farms in an enclosed bay may degrade water quality and thus impede eelgrass growth throughout the bay, even if the netpens are not situated directly over eelgrass beds). Consequence depends on the effects of activities on habitat area and density, and the ability of habitats to recover from these effects (i.e. through processes such as recruitment and regeneration). Outputs from the model are useful for understanding the relative risk of human activities and climate change to habitats within a study region and among alternative future scenarios. Model outputs can help identify areas on the seascape where human activities may create trade-offs among environmental services by posing risk high enough to compromise habitat structure and function. The model can help to prioritize areas for conservation and inform the design and configuration of spatial plans for both marine and terrestrial systems. This is a Tier 0 model.
11.2 Introduction
Nearshore habitats such as kelp forests and eelgrass meadows provide valuable environmental services including the protection of shorelines from storms, nursery habitat for fisheries and carbon storage and sequestration. As these habitats become degraded by human activities, the environmental services they provide are threatened. The impacts of human activities in coastal areas, both on land and in the sea, are pervasive in coastal ecosystems. Recent global analyses have revealed that almost no area of the world’s oceans is untouched by human impacts (Halpern et al. 2008). Thus, an understanding of the location and intensity of human impacts on nearshore ecosystems is an essential component of informed and successful coastal and ocean management. The InVEST HRA model allows users to assess the threat of human activities to the health of these ecosystems.
170 InVEST User Guide, Release 2.5.6
11.2.1 InVEST biodiversity model vs. InVEST habitat risk assessment model
A primary goal of conservation is the protection of biodiversity; biodiversity is intricately linked to the production of environmental services. While some consider biodiversity itself to be an environmental service, the InVEST biodi- versity model (found in the terrestrial toolbox) treats it as an independent attribute of natural systems, with its own intrinsic value (InVEST does not monetize biodiversity). InVEST includes a biodiversity model because natural re- source managers, corporations and conservation organizations are becoming increasingly interested in understanding how and where biodiversity and environmental services align in space and how management actions affect both. The biodiversity model uses habitat quality and rarity as a proxy for diversity. When developing a similar model with marine systems in mind, differences in data availability (e.g. the rarity of an analog to land-use/land-cover maps in marine systems) and differences in thinking (e.g. the prevalence of a risk- assessment framework in fisheries science) led us to the development of the habitat risk assessment model described in this chapter. Both the biodiversity model and the HRA model can be used to identify areas on a landscape or seascape where the risk posed by human activities is highest. Indeed, while the two models are segregated into the marine and terrestrial toolboxes, they can be used across systems. However, the modeling approaches differ in several ways. First, the exposure-consequence framework of the HRA model allows model results to be interpreted in a manner that helps users explore which types of management strategies are likely to most effectively reduce risk (Figure 1). For example, ecosystems with high exposure and high consequence may be targeted for intense active management, while effective strategies for ecosystems with low exposure to human stressors but high consequence may include close monitoring but little active intervention unless exposure increases. Second, the transparent flexible structure and explicit visualization of data uncertainty in the HRA model facilitate its use in both data-rich and data-poor situations. Finally, the biodiversity model is better suited for terrestrial applications than marine applications because it requires a land use land cover map as an input. The HRA model can be used in both marine and terrestrial systems.
Figure 11.1: Habitats with high exposure to human activities and high consequence are at high risk. Plotting exposure and consequence data in this plot allows users to visualize risk, and to assess which types of risk are more effectively mitigated by human intervention (risks driven by exogenous human factors, top right region of the risk space) and which types of risk are better addressed through monitoring and preparedness (risks driven by endogenous habitat- specific factors). (Adapted from Dawson et al. 2011).
Risk assessment has a long history in the field of ecotoxicology, and is now emerging as a valuable method in
11.2. Introduction 171 InVEST User Guide, Release 2.5.6 ecosystem-based fisheries management (Astles et al. 2006, Hobday et al. 2011). Risk assessment is used to determine the likelihood that a hazard will cause undesired consequences (Burgman 2005). In the context of marine ecosystem- based management, risk assessment evaluates the probability that human activities will impede the achievement of desired marine management objectives. In the HRA model, we define risk as the likelihood that human activities will reduce the quality of nearshore habitats to the point where their ability to deliver environmental services is impaired. Researchers have made significant progress in evaluating human impacts on marine ecosystems in recent years. How- ever many of these approaches lack generality because they are focused on the effects of a single sector (i.e. fisheries e.g. Astles et al. 2006, Hobday et al. 2011), or have limited transparency and flexibility because they are based on expert opinion (Halpern et al. 2008, Teck et al. 2010). The HRA model in Marine InVEST builds on these approaches and allows users to evaluate the risk posed by a variety of human activities to key coastal habitats in a transparent, repeatable and flexible way.
11.3 The model
The risk of human activities to coastal and nearshore habitats is a function of the habitat’s exposure to the activity and the consequence of exposure. To determine exposure, users provide model inputs such as base maps of habitat distribution and human activities, the timing and intensity of the activity and the effectiveness of current management practices in safeguarding habitats. To determine consequence, users provide model inputs such as observed loss of habitat and the ability of habitats to recover. The model is flexible and can accommodate data-poor and data-rich situations. Data may come from a combination of peer-reviewed sources at the global scale and locally available fine-scale data sources. Model inputs and results can be updated as better information becomes available. The HRA model produces maps that display the relative risk from a variety of human activities on habitats within a study region and among alternative future scenarios. When run as part of a complete Marine InVEST analysis, the HRA model can be used to identify which human activities are likely to cause trade-offs in other environmental services. As a result, the model will help managers prioritize and evaluate management strategies with regards to their effectiveness of reducing risks to nearshore habitats and maintaining the delivery of desired environmental services.
11.3.1 How it works
The HRA model combines information about the exposure of habitats to each stressor with information about the consequence of that exposure for each habitat to produce maps of risk to habitats and habitat quality. By default, exposure depends on the extent of geographic overlap between habitats and human activities, the duration of time that the activity and habitat overlap, the intensity of the stressor and the degree to which management strategies mitigate impact. The default consequence scoring depends on the degree of habitat loss, change in habitat structure and the ability of habitats to recover from these effects (i.e., through life history traits such as recruitment and regeneration rates). The modelling approach is flexible so if any of the factors that influence exposure and consequence are irrelevant in a particular case, they need not be included (see Risk of human activities to habitats).
Risk of human activities to habitats
The risk of human activities to habitats is modeled in four steps. Step 1. The first step involves determining the likelihood of exposure of the habitat to the stressor and the consequence of this exposure. Exposure and consequence are both determined by assigning a rating (typically 0-3) to a set of criteria for each attribute. We have provided the user with a set of standard used criteria, but any criteria may be added or removed. Guidelines for scoring of the default criterion are summarized below, and abbreviated descriptions of scoring on a 0-3 basis are provided in the tables produced from HRA Preprocessor. To ensure accuracy, we recommend that scores be determined using readily available data from peer-reviewed literature or published reports, however, you are free to use any data you believe to be the most accurate. For each score assigned, you should also indicate the quality of the data used to determine the score and the weighted importance of the criteria relative to other criteria. This allows you to assign greater weight to criteria where scoring confidence was higher, or to criteria which contribute more to
11.3. The model 172 InVEST User Guide, Release 2.5.6
risk in the system. Thus, the overall exposure E and consequence C scores are calculated as weighted averages of the exposure values ei and consequence values ci for each criterion i as
PN ei E = i=1 di·wi (11.1) PN 1 i=1 di·wi
PN ci C = i=1 di·wi (11.2) PN 1 i=1 di·wi
where di represents the data quality rating for criterion i, wi represents the importance weighing for criterion i and N is the number of criteria evaluated for each habitat. Step 2. The second step combines the exposure and response values to produce a risk value for each stressor-habitat combination. There are two options for risk calculation. For Euclidean Risk calculation, risk to habitat i caused by stressor j is calculated as the Euclidean distance from the origin in the exposure-consequence space,
p 2 2 Rij = (E − 1) + (C − 1) (11.3)
For Multiplicative Risk calculation, risk to habitat i caused by stressor j is calculated as the product of the summed exposure and consequence scores.
Rij = E · C (11.4)
11.3. The model 173 InVEST User Guide, Release 2.5.6
Step 3. In this step, the model quantifies the cumulative risk of all stressors on the habitats. Cumulative risk for habitat i is the sum of all risk scores for each habitat,
J X Ri = Rij (11.5) j=1
Step 4. In the final step, the model identifies areas of habitats that are risk ‘hotspots’. These are areas where the influence of human-derived stressors is so great that ecosystem structure and function may be severly compromised. In these areas, there may be trade-offs between human activities and a range of ecosystem services. Thus, users may choose to consider these habitats to be functionally absent in inputs to other InVEST ecosystem service models (see the Interpreting Results section for guidance on how to use risk hotspots to identify trade-offs among human activities under alternative scenarios). Habitat area is classified as HIGH, MED or LOW risk based on risk posed by any individual stressor and the risk posed by the cumulative effects of multiple stressors. A classification of HIGH is assigned to area that has a risk of >66% of the maximum risk score for any individual stressor, or >66% of total possible cumulative risk. Cells are classified as MED if they have individual stressor or cumulative risk scores between 33%-66% of the maximum score. Cells are classified as LOW risk if they have individual or cumulative risk scores of 0-33% of the maximum score.
Exposure and consequence criteria in more detail
The model allows for any number of criteria to be used when evaluating the risk to habitat areas. Below are provided a number of typical considerations when evaluating the role that stressors will have on a habitat. These criteria are rated on a scale of 0-3, but evaluation of relevance is not constrained to these numbers. If there is significant literature using an alternative scale, this can also be used for rating criteria as long as there is consistency across the rating scores within the single model run. It should be noted that using a score of 0 on ANY scale will indicate that the given criteria is not desired within that model run.
Exposure of habitats to stressors
The risk of a habitat being affected by a stressor depends in part on the exposure of the habitat to that stressor. Stressors may impact habitats directly and indirectly. Because indirect impacts are poorly understood and difficult to trace, we only model the risk of stressors that directly impact habitat by overlapping in space. Other important considerations include the duration of spatial overlap, intensity of the stressor, and whether management strategies reduce or enhance exposure. 1. Spatial overlap . To assess spatial overlap in the study area, the model uses maps of the distribution of habitats and stressors. Habitat types can be biotic, such as eelgrass or kelp, or abiotic, such as hard or soft bottom. The user defines the detail of habitat classification. For example, habitats can be defined as biotic or abiotic, by taxa (e.g., coral, seagrass, mangrove), by species (e.g., red, black mangroves) or in whatever scheme the user desires. However, the user should keep in mind that in order for such detail to be useful and change the outcome of the model, these habitat classifications should correspond with differences between habitats in their response to the stressors. Additionally, the model requires the user to input maps of the distribution and zone of influence of each stressor. The zone of influence of each stressor is the distance over which the stressor spreads beyond the footprint of the stressor indicated in the stressor map. For some stressors, such as over-water structures that shade eelgrass beds, this distance will be small. For other stressors, such as finfish aquaculture pens where nutrients spread 300-500m this distance may be large. The model uses the maps of habitat and stressor distributions to calculate the amount of each habitat type that overlaps with each stressor and its zone of influence. 2. Overlap time rating. Temporal overlap is the duration of time that the habitat and the stressor experience spatial overlap. Some stressors, such as permanent overwater structures, are present year-round; others are seasonal,
11.3. The model 174 InVEST User Guide, Release 2.5.6
such as certain fishing practices. Similarly, some habitats (e.g. mangroves) are present year round, while others are more ephemeral (e.g. some seagrasses). We use the following categories to classify HIGH, MEDIUM and LOW temporal overlap: High (3) Medium (2) Low (1) No score (0) Tempo- Habitat and stressor Habitat and stressor Habitat and stressor N/A ral co-occur for 8-12 months co-occur for 4-8 months co-occur for 0-4 months overlap of the year of the year of the year Choose “No score” to exclude this criteria from your assessment. 3. Intensity rating. The exposure of a habitat to a stressor depends not only on whether the habitat and stressor overlap in space and time, but also on the intensity of the stressor. The intensity criterion is stressor-specific. For example, the intensity of nutrient-loading stress associated with netpen salmon aquaculture is related to the number of salmon in the farm and how much waste is released into the surrounding environment. Alternatively, the intensity of destructive shellfish harvesting is related to the number of harvesters and the harvest practices. You can use this intensity criteria to explore how changes in the intensity of one stressor might affect risk to habitats. For example, one could change the intensity score to represent changes in the stocking density of a salmon farm in a future scenario. One can also use this ranking to incorporate relative differences in the intensity of different stressors within the study region. For example, different types of marine transportation may have different levels of intensity. For example, cruise ships may be a more intense stressor than water taxis because they release more pollutants than the taxis do. We use the following categories to classify HIGH, MEDIUM and LOW intensity: High (3) Medium (2) Low (1) No score (0) Intensity High intensity Medium intensity Low intensity N/A Choose “No score” to exclude this criteria from your assessment. 4. Management strategy effectiveness rating. Management can limit the negative impacts of human activities on habitats. For example, policies that require salmon aquaculturists to let their farms lie fallow may reduce the amount of waste released and allow nearby seagrasses to recover. Similarly, regulations that require a minimum height for overwater structures reduce the shading impacts of overwater structures on submerged aquatic vegetation. Thus, effective management strategies will reduce the exposure of habitats to stressors. The effectiveness of management of each stressor is scored relative to other stressors in the region. So if there is a stressor that is very well managed such that it imparts much less stress on the system than other stressors, classify management effectiveness as “very effective.” In general, however, the management of most stressors is likely to be “not effective.” After all, you are including them as stressors because they are having some impact on habitats. You can then use this criterion to explore changes in management between scenarios, such as the effect of changing coastal development from high impact (which might receive a score of “not effective”) to low impact (which might receive a score of “somewhat effective).” We use the following categories to classify HIGH, MEDIUM and LOW management effectiveness: High (3) Medium (2) Low (1) No score (0) Management effectiveness Not effective, poorly managed Somewhat effective Very effective N/A Choose “No score” to exclude this criteria from your assessment.
Consequence of exposure
The risk of a habitat being degraded by a stressor depends on the consequence of exposure. The consequence of exposure depends on the ability of a habitat to resist the stressor and recover following exposure, and can be assessed
11.3. The model 175 InVEST User Guide, Release 2.5.6 using four key attributes: change in area, change in structure, frequency of natural disturbance, and recovery attributes. We describe each in turn below. 1. Change in area rating. Change in area is measured as the percent change in areal extent of a habitat when exposed to a given stressor and thus reflects the sensitivity of the habitat to the stressor. Habitats that lose a high percentage of their areal extent when exposed to a given stressor are highly sensitive, while those habitats that lose little area are less sensitive and more resistant. We use the following categories to classify HIGH, MEDIUM and LOW change in area: High (3) Medium (2) Low (1) No score (0) Change in High loss in area Medium loss in area Low loss in area N/A area (50-100%) (20-50%) (0-20%) Choose “No score” to exclude this criteria from your assessment. 2. Change in structure rating. For biotic habitats, the change in structure is the percentage change in structural density of the habitat when exposed to a given stressor. For example, change in structure would be the change in shoot density for seagrass systems, change in polyp density for corals, or change in stipe density for kelp systems. Habitats that lose a high percentage of their structure when exposed to a given stressor are highly sensitive, while habitats that lose little structure are less sensitive and more resistant. For abiotic habitats, the change in structure is the amount of structural damage sustained by the habitat. Sensitive abiotic habitats will sustain complete or partial damage, while those that sustain little to no damage are more resistant. For example, gravel or muddy bottoms will sustain partial or complete damage from bottom trawling while hard bedrock bottoms will sustain little to no damage. We use the following categories to classify HIGH, MEDIUM and LOW change in structure: High (3) Medium (2) Low (1) No score (0) Change High loss in structure (for Medium loss in structure (for Low loss in structure (for N/A in biotic habitats, 50-100% biotic habitats, 20-50% loss biotic habitats, 0-20% loss in struc- loss in density, for abiotic in density, for abiotic density, for abiotic habitats, ture habitats, total structural habitats, partial structural little to no structural damage) damage) damage) Choose “No score” to exclude this criteria from your assessment. 3. Frequency of natural disturbance rating. If a habitat is naturally frequently perturbed in a way similar to the anthropogenic stressor, it may be more resistant to additional anthropogenic stress. For example, habitats in areas that experience periodical delivery of nutrient subsidies (i.e. from upwelling or allocthonous inputs such as delivery of intertidal plant material to subtidal communities) are adapted to variable nutrient conditions and may be more resistant to nutrient loading from netpen salmon aquaculture. This criterion is scored separately for each habitat-stressor combination, such that being adapted to variable nutrient conditions increases resistance to nutrient loading from salmon aquaculture but not destructive fishing. However, high storm frequency may increase resistance to destructive fishing, because both stressors impact habitats in similar ways. We use the following categories to classify HIGH, MEDIUM and LOW natural disturbance frequencies: High (3) Medium (2) Low (1) No score (0) Frequency of natural disturbance Annually or less often Several times per year Daily to weekly N/A Choose “No score” to exclude this criteria from your assessment.
Note: The following consequence criteria are Recovery Attributes. These include life history traits such as regenera- tion rates and recruitment patterns influence the ability of habitats to recover from disturbance. For biotic habitats, we treat recovery as a function of natural mortality, recruitment, age of maturity, and connectivity.
11.3. The model 176 InVEST User Guide, Release 2.5.6
4. Natural mortality rate rating (biotic habitats only). Habitats with high natural mortality rates are generally more productive and more capable of recovery. We use the following categories to classify HIGH, MEDIUM and LOW natural mortality rates: High (3) Medium (2) Low (1) No score (0) Natural Low mortality (e.g. Moderate mortality (e.g. High mortality (e.g.80% N/A mortality rate 0-20%) 20-50%) or higher) Choose “No score” to exclude this criteria from your assessment. 5. Recruitment rating (biotic habitats only). Frequent recruitment increases recovery potential by increasing the chance that incoming propagules can re-establish a population in a disturbed area. We use the following categories to classify HIGH, MEDIUM and LOW natural recruitment rate: High (3) Medium (2) Low (1) No score (0) Natural recruitment rate Every 2+ yrs Every 1-2 yrs Annual or more often N/A Choose “No score” to exclude this criteria from your assessment. 6. Age at maturity/recovery time. Biotic habitats that reach maturity earlier are likely to be able to recover more quickly from disturbance than those that take longer to reach maturity. Here we refer to maturity of the habitat as a whole (i.e., a mature kelp forest) rather than reproductive maturity of individuals. For abiotic habitats, shorter recovery times for habitats such as mudflats decrease the consequences of exposure to human activities. In contrast, habitats made of bedrock will only recover on geological time scales, greatly increasing the consequences of exposure. We use the following categories to classify HIGH, MEDIUM and LOW age at maturity/recovery time: High (3) Medium (2) Low (1) No score (0) Age at maturity/recovery time More than 10 yrs 1-10yrs Less than 1 yr N/A Choose “No score” to exclude this criteria from your assessment. 7. Connectivity rating (biotic habitats only). Larval dispersal and close spacing of habitat patches increases the recovery potential of a habitat by increasing the chance that incoming propagules can re-establish a population in a disturbed area. We use the following categories to classify HIGH, MEDIUM and LOW connectivity: High (3) Medium (2) Low (1) No score (0) Connec- Low dispersal (less than Medium dispersal High dispersal N/A tivity 10km) (10-100km) (>100km) Choose “No score” to exclude this criteria from your assessment.
Using Spatially Explicit Criteria
In addition to assigning a single rating to a criteria for the given influence on a habitat or stressor, the model also allows for spatially explicit criteria to be used as an input. These are vector layers, where each feature may contain a separate rating for that particular area. (See the HRA Preprocessor section for more information how how to prepare and use spatially explicit criteria within a complete model run.)
Guidelines for scoring data quality
Risk assessment is an integrative process, which requires a substantial amount of data on many attributes of human and ecological systems. It is likely that some aspects of the risk assessment will be supported by high quality data and
11.3. The model 177 InVEST User Guide, Release 2.5.6 other aspects will be subject to limited data availability and high uncertainty. The user has the option of scoring data quality to put greater weight on the criteria for which confidence is higher in the calculation of risk (eq. 2 and 3). We hope that by including the option to rate data quality in the model, users will be aware of some sources of uncertainty in the risk assessment, and will therefore be cautious when using results derived from low quality data. In addition, the information generated from this rating process can be used to guide research and monitoring effects to improve data quality and availability. If users do not have verified information on the data quality of a given criteria, they should use a 1 to indicate the lowest possible data quality score. For each exposure and consequence score, users can indicate the quality of the data that were used to determine the score on a sliding scale where 1 indicates a limited knowledge of the data quality, and anything above that would be seen as increasingly trustworthy data. Best data Adequate data Limited data Substantial information is available Information is based on data collected No empirical literature exists to support the score and is based on outside the study region, may be based to justify scoring for the data collected in the study region (or on related species, may represent species but a reasonable nearby) for the species in question. moderate or insignificant statistical inference can be made by the relationships. user.
11.3.2 Limitations and Assumptions
Limitations
1. Results are limited by data quality: The accuracy of the model results is limited by the availability and quality of input data. Using high quality data such as those from local assessments replicated at several sites within the study region for the species in question within the last ten years will yield more accurate results than using lower quality data that are collected at a distant location with limited spatial or temporal coverage. In most cases, users will need to use data from other geographic locations for some of the stressor-habitat combinations because most of the data on the effects of some stressors have only been collected in a limited number of locations worldwide. To overcome these data limitations, we include a data quality score in the analysis. This score allows users to down-weight criteria for which data quality is low. 2. Results should be interpreted on a relative scale: Due to the nature of the scoring process, results can be used to compare the risk of several human activities among several habitats within the study region (which can range in size from small local scales to a global scale), but should not be used to compare risk calculations from separate analyses. 3. Results do not reflect the effects of past human activities. The HRA model does not explicitly account for the effects of historical human activities on the current risk. Exposure to human activities in the past may affect the consequence of human activities in the present and future. If users have historical data on the exposure of habitats to human activities (e.g. spatial and temporal extent), and information on how this affects current consequence scores, they may include this information in the analysis for more accurate results. 4. Results are based on equal weighting of criteria unless the user weights the criteria by importance or data quality. The model calculates the exposure and consequence scores assuming that the effect of each criterion (i.e. spatial overlap and recruitment pattern) is of equal importance in determining risk. The relative importance of each of the criteria is poorly understood, so we assume equal importance. However, the user has the option to weight the importance of each criterion in determining overall risk.
Assumptions
1. Often information in the literature about the effect stressors on habitats comes from only a few locations. If using globally available data or data from other locations, users make the assumption that ecosystems around the world respond in similar ways to any given stressor (i.e. eelgrass in the Mediterranean responds to netpen
11.3. The model 178 InVEST User Guide, Release 2.5.6
aquaculture in the same way as eelgrass in British Columbia). To avoid making this assumption across the board, users should use local data whenever possible. 2. Cumulative risk is additive (vs. synergistic or antagonistic). The interaction of multiple stressors on marine ecosystems is poorly understood (see Crain et al. 2008 for more information). Interactions may be additive, synergistic or antagonistic. However, our ability to predict the type of interaction that will occur is limited. Due to the absence of reliable information on the conditions that determine additivity, synergism or antagonism, the model assumes additivity because it is the simplest approach. As a result, the model may over- or under-estimate the cumulative risk depending on the set of stressors occurring in the study region.
11.4 Data Needs
The model uses an interface to input all required and optional data and a series of Comma Separated Value (CSV) files with which to score all criteria and their data quality. Here we outline the options presented to the user via the interface and the maps and data tables that will be used by the model. First we describe required inputs, followed by a description of optional inputs. To run the model, three steps are required: 1. Run the HRA Preprocessor Tool 2. Fill out the Ratings CSVs 3. Run the Habitat Risk Assessment model
11.4.1 HRA Preprocessor
Before running the HRA model, it is necessary to concatenate and rate all applicable criteria information. This can be accomplished by running the Preprocessor tool, then editing the resulting CVSs. If you have already run the model, or have the ‘habitat_stressor_ratings’ directory from a previous HRA Preprocessor run, you may skip this step and proceed to running the Habitat Risk Assessment tool. To run the tool, run the HRA Preprocessor executable. This will launch a graphical user interface (GUI).
There are several pieces that should be used as inputs to this tool. At any time, you can click the blue question marks to the right of an input for additional guidance. 1. Workspace Location (required). Users are required to specify a workspace folder path. Running HRA Prepro- cessor creates a folder named ‘habitat_stressor_ratings’ within this workspace. This Folder will hold all relevant CSVs for criteria rating in the particular model run. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\HabitatRiskAssess_3_0\
2. Calculate Risk To Habitat/Species Here you will select the habitats and/or species that will be inputs for this run of the model. Each of these inputs should point to a directory containing all of the named habitat or species shapefile layers that you wish to include in this model run. The file names are not required to contain an identifying number. Each directory should be independent of the others so as to avoid incorrect repetition in the outputs, and should contain ONLY layers that are desired within this assessment. Name: Path to a habitat or species folder. Avoid spaces. Sample path: \InVEST\HabitatRiskAssess_3_0\HabitatLayers
11.4. Data Needs 179 InVEST User Guide, Release 2.5.6
Figure 11.2: The HRA Preprocessor main user interface.
11.4. Data Needs 180 InVEST User Guide, Release 2.5.6
3. Directory for Stressor Layers Users should select a folder containing stressors to be overlapped with habitats and/or species. This directory should contain ONLY the stressors desired within this model run. Name: Path to a habitat or species folder. Avoid spaces. Sample path: \InVEST\HabitatRiskAssess_3_0\HabitatLayers
4. Criteria We have divided up criteria into 3 categories: Exposure, Sensitivity, and Resilience. Exposure crite- ria are specific to a habitat-stressor pairing, and will be applied to the exposure portion of the risk modeling equation. Sensitivity criteria are also applied to a specific habitat-stressor pairing, but will be applied to the consequence portion of the risk equation. Resilience criteria will likewise be applied to the consequence portion of the risk equation, but are specific to an overall habitat. Placing a criteria into one of these categories within the user interface will determine how user ratings are input into the HRA model. The default criteria provided are derived from peer-reviewed literature and are recommended as a good set of contributers to risk in a system, but users do have the option to add or remove criteria if desired. Only choose this option if the default criteria do not apply to the system being modeled, or do not correctly address all facets of the risk assessment.
5. Optional Determine whether spatial criteria are available and desired. These are vector layer files which would provide more explicit detail for a specific criteria in the assessment. If spatially explicit criteria is desired, this input should point to an outer directory for all spatial criteria. A rigid structure MUST be followed in order for the model to run. Within the outer spatial criteria folder, there MUST be the following 3 folders: Sensitivity, Exposure, and Resilience. Vector criteria may then be placed within the desired folder. Each feature in the shapefiles used MUST include a ‘Rating’ attribute which maps to a float or int value desired for use as the rating value of that spatial criteria area. • Any criteria placed within the Resilience folder will apply only to a given habitat. They should be named with the form: habitatname_criteria_name.shp. Criteria may contain more than one word if separated by an underscore. • Any criteria placed within the Exposure folder will apply to the overlap between a given habitat and a given stressor. They should be named with the form: habitatname_stressorname_criteria_name.shp. Criteria may contain more than one word if separated by an underscore. • Any criteria placed within the Sensitivity folder will apply to the overlap between a given habitat and a given stressor. They should be named with the form: habitatname_stressorname_criteria_name.shp. Criteria may contain more than one word if separated by an underscore. 6. Run the tool. This will create a directory in your selected workspace called habitat_stressor_ratings. Keep in mind that if a folder of the name habitat_stressor_ratings already exists within the workspace, it will be deleted to make way for the new output folder. This directory can be renamed as necessary after completion, and will contain a series of files with the form: habitatname_ratings.csv, as well a file named stressor_buffers.csv. There will be one file for every habitat, and the one additional file for stressor buffers. HRA 3.0’s sample data includes a sample folder for use within the main HRA executable called habitat_stressor_ratings_sample, containing pre-filled criteria values relevant to the sample data for the west coast of Vancouver Island, Canada.
11.4. Data Needs 181 InVEST User Guide, Release 2.5.6
Figure 11.3: The three categories- Exposure, Sensitivity, and Resilience correspond to stressor-specific, overlap- specific, and habitat-specific criteria respectively.
11.4. Data Needs 182 InVEST User Guide, Release 2.5.6
11.4.2 Ratings CSVs
The CSVs contained within the habitat_stressor_ratings folder will provide all criteria information for the run of the Habitat Risk Assessment. There are two types of CSVs- habitat overlap CSVs and the stressor buffer CSV. Habitat CSVs will contain not only habitat-specific criteria information, but also all criteria that impact the overlap between that habitat and all applicable stressors. The stressor buffer CSV will be a single file containing the desired buffer for all stressors included in the assessment.
Figure 11.4: Upon initial creation, CSVs will contain no ratings, only guidance for known criteria on a scale of 0-3. However, users should feel free to fill in ratings on a different scale if there is significant reviewed data, but should be sure to be consistent on scale across ALL CSVs.
When preprocessor is run, the CSVs will contain no numerical ratings, only guidance on how each rating might be filled out. The user should use the best available data sources in order to obtain rating information. The column
11.4. Data Needs 183 InVEST User Guide, Release 2.5.6 information to be filled out includes the following: 1. “Rating”- This is a measure of a criterion’s impact on a particular habitat or stressor, with regards to the overall ecosystem. Data may come from a combination of peer-reviewed sources at the global scale and locally available fine-scale data sources. Model inputs and results can be updated as better information becomes available. We provide guidance for well-known criteria on a scale of 0-3, but it should be noted that if information is available on a different scale, this can also be used. It is important to note, however, that all rating information across all CSVs should be on one consistent scale, regardless of what the upper bound is. 2. “DQ”- This column represents the data quality of the score provided in the ‘Rating’ column. Here the model gives the user a chance to downweight less-reliable data sources, or upweight particularly well-studied criteria. While we provide guidance for a scoring system of 1-3, the user should feel free to use any upper bound they feel practical, as long as the scale is consistent. The lower bound, however, should ALWAYS be 1, unless the user wishes to remove the entire criteria score. 3. “Weight”- Here the user is given the opportunity to upweight critiera which they feel are particularly important to the system, independent of the source data quality. While we provide guidance for a scoring system from 1-3, the user should feel free to use any upper bound they feel practical, as long as the scale is consistent. The lower bound, however, should ALWAYS be 1 unless the user wishes to remove the entire criteria score. 4. (Optional) “E/C”- This column indicates whether the given criteria is being applied to the exposure or the consequence portion of the chosen risk equation. These can be manually changed by the user on a single criteria basis, however, we would strongly recommend against it. If the user desires to chance that criteria’s allocation, it would be better to chance the allocation of the criteria within the Resilience, Exposure, Sensitivity categories using the HRA Preprocessor User Interface. By default, any criteria in the Sensitivity or Resilience categories will be assigned to Consequence (C) within the risk equations, and any criteria within the Exposure category will be assigned to Exposure (E) within the risk equation. Habitat CSVs should be filled out with habitat-specific criteria information as well as any criteria which apply to the overlap of the given habitat and stressors. The Stressor Buffer CSV should be filled out with the desired numerical buffer which can be used to expand a given stressor’s influence within the model run. This can be 0 if no buffering is desired for a given stressor, but may NOT be left blank. Any criteria which use spatially explicit criteria (specified by the user during the HRA Preprocessor) will be noted in the CSV by the word ‘SHAPE’ in the rating column for that habitat, stressor, or combined criteria. The user should still fill in a Data Quality and Weight for these criteria, but should NOT remove the ‘SHAPE’ string unless they no longer desire to use a spatial criteria for that attribute.
11.4. Data Needs 184 InVEST User Guide, Release 2.5.6
11.4.3 Habitat Risk Assessment
The main computation portion of the HRA model will be done by the Habitat Risk Assessment executable. First we describe required inputs. The required inputs are the minimum data needed to run this model.
Figure 11.5: The HRA 3.0 main executable.
1. Workspace Location (required). Users are required to specify a workspace folder path. It is recommended that the user create a new folder for each run of the model. For example, by creating a folder called “runBC” within the “HabitatRiskAssess_3_0” folder, the model will create “Intermediate” and “Output” folders within this “runBC” workspace. The “Intermediate” folder will compartmentalize data from intermediate processes. The model’s final outputs will be stored in the “Output” folder. Name: Path to a workspace folder. Avoid spaces. Sample path: \InVEST\HabitatRiskAssess_3_0\runBC
2. Criteria Scores Folder (required). After running the HRA Preprocessor tool, a folder will be created which contains the collective criteria scores for all habitats and stressors. For this input, point to the outer folder containing all CSVs.
11.4. Data Needs 185 InVEST User Guide, Release 2.5.6
Name: Folder can be named anything, but avoid spaces. Sample path: \InVEST\HabitatRiskAssess_3_0\runBC\habitat_stressor_ratings
3. Resolution of Analysis (required). The size in meters that is desired for the analysis of the shapefile layers. This will define the width and height of each unique risk grid cell. This must be a whole number. 4. Risk Equation (required). This selection chooses the equation that will be used when calculating risk to a given habitat. (See the Risk of human activities to habitats section.) The user may choose either either a Euclidean risk model, or a Multiplicative risk model. 5. Decay Equation (required) This selection picks how any desired stressor buffering will be applied. The stressor buffer amount, which should have been provided by the user in the stressor buffer CSV, can be degraded in order to provide a more accurate depiction of influence decay within an ecosystem. The options are as follows. “None” will provide the full range of the stressor buffer, without any decay whatsoever. “Linear” and “Exponential” will use the stated equation as a model for decay from the edges of the original stressor shape to the full extent of the buffer distance. 6. Maximum Criteria Score (required) The maximum criteria score is the user-reported highest value assigned to any criteria rating within the assessment. This will be used as the upper bounded value against which all rating scores will be compared. For example, in a model run where the ratings scores vary from 0-3, this would be a 3. If the user chooses to use a different scale for ratings, however, this should be the highest value that could be potentially assigned to a criteria. If the model run is using spatially explicit criteria, this value should be the maximum value assigned to either a criteria feature or to a CSV criteria rating. 7. Use Subregions Shapefile? (optional). By checking this box, the model will use a subregions shapefile to generate an HTML table of averaged exposure, consequence, and risk values within each subregion by habitat and stressor. In addition, if the Risk Equation chosen is Euclidean, the model will also generate a series of figures which clearly display the exposure-consequence ratings and the resulting risk results for each habitat-stressor combination by subregion. It will also create a figure showing cumulative ecosystem risk for all subregions habitats in the study. Each of the subregion shapefile features MUST contain a ‘Name’ attribute in order to be properly included in the subregion averaging. Name: File can be named anything, but avoid spaces. File Type: Polygon shapefile (.shp) Sample path: \InVEST\HabitatRiskAssess_3_0\runBC\subregions.shp
11.5 Interpreting results
11.5.1 Model outputs
Upon successful completion of the model, you will see new folders in your Workspace called “Intermediate” and “Output”. These two folders will hold all outputs, both temporary and final that are used in a complete run of the model. While most users will be interested only in the Output folder data, we will describe all outputs below.
Intermediate Folder
The Intermediate folder contains files that were used for final output calculations. All rasters within this file use the pixel size that the user specifies in the “Resolution of Analysis” text field of the Habitat Risk Assessment main executable. • \Intermediate\Criteria_Rasters\spatial_criteria_name.tif – If the user has included any spatially explicit criteria in the assessment, this folder will contain a rasterized version of that vector layer, with the ‘Rating’ attribute burned as the pixel value.
11.5. Interpreting results 186 InVEST User Guide, Release 2.5.6
• \Intermediate\Habitat_Rasters\habitat_name.tif – A rasterized version of all habitat or species vector files included in the assessment. • \Intermediate\Stressor_Rasters\stressor_name.tif – A rasterized version of all stressor vector files included in the assessment. • \Intermediate\Stressor_Rasters\stressor_name_buff.tif – This is a copy of the stressor_name.tif file in the same folder, but with each stressor’s individual buffering included. If a given stressor has a 0 buffer distance, this will be an exact copy of the rasterized vector file. For all other files, this will be buffered by the desired amount set forth in the “Stressor Buffer (m)” section of the Ratings CSVs, decayed from 1 to 0 using the equation chosen in the “Decay Equation” section of the Habitat Risk Assessment. • \Intermediate\Overlap_Rasters\H[habitat_name]_S[stressor_name].tif – A raster representing the overlap between each pair of the habitat or species rasters, and the buffered stressor rasters. • \Intermediate\H[habitatname]_S[stressorname]_C_Risk_Raster.tif – A raster representing the Consequence portion of the final risk calculations for the overlap of the given habitat and stressor. • \Intermediate\H[habitatname]_S[stressorname]_E_Risk_Raster.tif – A raster representing the Exposure portion of the final risk calculations for the overlap of the given habitat and stressor. • Intermediate\H[habitatname]_S[stressorname]_Risk_Raster.tif – A raster containing the final risk calculation for the given habitat and stressor combination. This risk raster takes into account each of the criteria that apply to the habitat and stressor, as well as the user-specified risk equation.
Output folder
The following is a short description of each of the final outputs from the HRA model. Each of these output files is saved in the “Output” folder that is saved within the user-specified workspace directory:
GIS
• \Output\maps\recov_potent_H[habitat_name].tif – This raster layer depicts the recovery potential of the each cell for the given habitat. Recovery potential is typically based on natural mortality rate, recruitment rate, age at maturity/recovery time and connectivity, though these can be altered by the user if alternate criteria are desired. Recovery potential is useful to those who are interested in identifying areas where habitats are more resilient to human stressors, and therefore may be able to withstand increasing stress. Habitats with low recovery potential are particularly vulnerable to intensifying human activities. • \Output\maps\ecosys_risk.tif – This raster layer depicts the sum of all cumulative risk scores for all habitats in each grid cell. It is best interpreted as an integrative index of risk across all habitats in a grid cell. For example, in a nearshore grid cell that contains some coral reef, mangrove and soft bottom habitat, the ecosys_risk value reflects the risk to all three habitats in the cell. The “ecosys_risk” value increases as the number of habitats in a cell exposed to stressors increases.
11.5. Interpreting results 187 InVEST User Guide, Release 2.5.6
• \Output\maps\cum_risk_H[habitat_name] – This raster layer depicts the cumulative risk for all the stressors in a grid cell on a habitat-by-habitat basis. For example, “cum_risk_eelgrass” depicts the risk from all stressors on habitat “eelgrass”. Cumulative risk is derived by summing the risk scores from each stressor (i.e. more stressors leads to higher cumulative risk). This layer is informative for users who want to know how cumulative risk for a given habitat varies across a study region (e.g. identify hotspots where eelgrass or kelp is at high risk from multiple stressors). Hotspots of high cumulative risk may be targeted for restoration or monitoring. • \Output\maps\H[habitat_name]_Risk.shp – These rasters are individual risk raster by habitat. These are the combination of all applicable habitat- stressor overlap rasters for this habitat summed into an overall risk raster. There is one habitat risk raster for each vector file originally used within the assessment. • \Output\maps\H[habitat_name]_LOW_RISK.shp – Shapefiles which represent only the LOW RISK areas of the given habitat. This is a polygonized version of the habitat raster. • Output\maps\H[habitat_name]_HIGH_RISK.shp – Shapefiles which represent only the LOW RISK areas of the given habitat. This is a polygonized version of the habitat raster.
HTML and plots
These outputs are optional, and their creation is dependent on user-provided subregion shapefiles. • \Output\HTML_Plots\risk_plot_AOI[aoiname].html – These figures show the cumulative risk for each habitat within a given subregion. There will be one subplot for every habitat or species. Within the habitat plot, there are points for every stressor. Each point is graphed by Exposure, Consequence values. If the risk equation chosen was Euclidean, the distance from the stressor point to the origin represents the average risk for that habitat, stressor pair within the selected AOI. Stressors that have high exposure scores and high consequence scores pose the greatest risk to habitats. Reducing risk through management is likely to be more effective in situations where high risk is driven by high exposure, not high consequence. • \Output\HTML_Plots\ecosystem_risk_plot.png – This figure shows the cumulative risk for each habitat in the study region by subregion. This figure can be used to determine which habitats are at highest risk from human activities, and if this risk is mostly due to high cumulative exposure (exogenous factors that can be mitigated by management) or high cumulative consequence (endogenous factors that are less responsive to human intervention). • \Output\HTML_Plots\Sub_Region_Averaged_Results_[yr-mon-day-min-sec].html – This HTML table is a concatenated set of data for all pairings of habitat and stressor within each provided subregion. For every pairing of habitat and stressor, the table provides average exposure, consequence, risk, and risk percentage (as a portion of total potential risk).
Log file
• hra-log-yr-mon-day-min-sec.txt – Each time the model is run a text file will appear in the workspace folder. The file will list the parameter values for that run and be named according to the date and time. – Parameter log information can be used to identify detailed configurations of each of scenario simulation.
11.5. Interpreting results 188 InVEST User Guide, Release 2.5.6
11.5. Interpreting results 189 InVEST User Guide, Release 2.5.6
11.6 References
Astles, K. L., Holloway, M. G., Steffe, A., Green, M., Ganassin, C., & Gibbs, P. J. 2006. An ecological method for qualitative risk assessment and its use in the management of fisheries in New South Wales, Australia. Fisheries Research, 82: 290-303. Burgman, M. 2005. Risks and decisions for conservation and environmental management. Cambridge University Press, Cambridge, UK. Crain, C. M., Kroeker, K., & Halpern, B. S. 2008. Interactive and cumulative effects of multiple human stressors in marine systems. Ecology Letters, 11: 1304-1315. Dawson, T. P., Jackson, S. T., House, J. I., Prentice, I. C., & Mace, G. M. 2011. Beyond Predictions: Biodiversity Conservation in a Changing Climate. Science, 332: 53-58. Halpern, B. S., Walbridge, S., Selkoe, K. A., Kappel, C. V., Micheli, F., D’Agrosa, C., Bruno, J. F., et al. 2008. A Global Map of Human Impact on Marine Ecosystems. Science, 319: 948-952. Hobday, A. J., Smith, A. D. M., Stobutzki, I. C., Bulman, C., Daley, R., Dambacher, J. M., Deng, R. A., et al. 2011. Ecological risk assessment for the effects of fishing. Fisheries Research, 108: 372-384. Teck, S. J., Halpern, B. S., Kappel, C. V., Micheli, F., Selkoe, K. A., Crain, C. M., Martone, R., et al. 2010. Using expert judgment to estimate marine ecosystem vulnerability in the California Current. Ecological Applications 20: 1402-1416. Williams, A., Dowdney, J., Smith, A. D. M., Hobday, A. J., & Fuller, M. 2011. Evaluating impacts of fishing on benthic habitats: A risk assessment framework applied to Australian fisheries. Fisheries Research, In Press.
11.6. References 190 CHAPTER TWELVE
MARINE WATER QUALITY MODEL: ADVECTION-DIFFUSION MODEL
12.1 Summary
Management of water quality plays a critical role in human and ecosystem health in coastal and estuarine ecosystems. The lack of a predictive understanding of the dispersal and fate of contaminants is a major obstacle to the develop- ment of management strategies for water quality problems. We developed a marine water quality model consisting of physical transport and biogeochemical processes to simulate the dispersal of water quality state variables (e.g. contam- inants) in response to changes in ecosystem structure driven by various management decisions and human activities. Hence, this model assesses how management and human activities influence the water quality in coastal and estuarine ecosystems. Although water quality is not an environmental service per se, the InVEST marine water quality model can be linked with other InVEST models to evaluate how changes in water quality might affect environmental services related to fisheries, aquaculture and recreation and how the exploitation of some services (e.g. aquaculture) might in turn affect water quality.
12.2 Introduction
The discharge of contaminants resulting from various management decisions and human activities may cause many types of water quality problems and potentially pose serious risks to both aquatic ecosystems and human health. There- fore, as human activities increase in coastal and marine ecosystems, water quality management has received increased attention in recent years. Since many processes (physical transport, biogeochemical and anthropogenic processes, etc.) affect water quality, it is difficult to determine the source of and to predict water quality problems. A numerical model based on physical and biogeochemical principals can help managers and decision makers investigate various water quality problems such as high concentrations of bacteria and toxic chemicals, hypoxia, and eutrophication (Park 1996). Contaminants introduced into an estuarine system are transported by water movement (i.e. physical transport) and, while being transported, their concentrations are modified by biogeochemical processes. Therefore physical and bio- geochemical processes combine to determine the fate of the contaminants. We developed a marine water quality model that accounts for both physical transport and biogeochemical processes to simulate the distribution and fate of a water quality state variable (e.g. contaminant or pollutant) in a coastal and estuarine system. The model allows users to change contaminant loadings from various sources, which may include sewage treatment plants, urban runoffs, storm sewers, failing septic systems, industrial discharges, floathomes, and aquaculture farms. For example, to explore the effects of alternate management schemes, users can alter pollutant or nutrient loading by adding, removing, or chang- ing practices at aquaculture farms. They can also define pollutant or nutrient loading due to land based management. The main output of the marine water quality model is a map of the concentration of a water quality state variable in response to the various management decisions under consideration. By exploring the concentration maps, users can
191 InVEST User Guide, Release 2.5.6
assess—in a spatially explicit manner—how management and development strategies influence the water quality in their target area. The marine water quality model can be linked with other InVEST models to evaluate other ecosystem services related to fisheries, aquaculture, habitat quality, and recreation.
12.3 The model
12.3.1 How it works
The marine water quality model calculates the spatial distribution of water quality state variables by solving a tidal- average horizontal two-dimensional mass-balance equation.
∂2C ∂2C ∂C ∂C ET + − U + V + S = 0 (12.1) ∂x2 ∂y2 ∂x ∂y
Where • x and y east and north coordinates, respectively • C tidal averaged concentration of a water quality state variable • U and V advective velocities (i.e., Eulerian residual current) in x and y directions, respectively • ET tidal dispersion coefficient • S term to account for sources and sinks of pollutant This is a steady state formulation of a classic advection diffusion equation. The first two terms on the left hand side represent tidal dispersion while the second two represent advective transport. The advective transport accounts for mass transport due to Eulerian residual current, which is obtained by averaging velocities at a fixed point over one or more tidal cycles. The tidal dispersion accounts for the dispersion of mass due to correlation between tidal components of velocity and concentration as well as the diffusion due to the turbulent fluctuations in velocity and concentration (MacCready & Geyer 2010). The tidal dispersion coefficient may be estimated by using observed salinity distribution or dye experiments. The observed tidal dispersion coefficient shows large variation ranging from 1 to 161 km2day−1 (Table 1) (Thomann & Mueller 1987). Note that in a future version of this model ET will be separated into a two dimensional vector with components Ex and Ey to correspond with each partial second order derivative in the first two terms of Equation (12.1). Table 1. Tidal dispersion coefficient (ET ) in various estuarine systems (modified from Table 3.3 in Thomann & Mueller 1987). Estuaries Tidal dispersion coefficient (km2day−1) Hudson River, NY 52 East River, NY 26 Wappinger and Fishkill Creek, NY 1-3 Delaware River, upper 5-18 Delaware River, lower 18-28 San Francisco Bay, southern 2-16 San Francisco Bay, northern 4-161 Rio Quayas, Ecuador 65 Thames River, England, low flow 5-7 Thames River, England, high flow 28
12.3. The model 192 InVEST User Guide, Release 2.5.6
Tidal dispersion coefficient may also be parametrized as a function of tidal flow and length scale of an estuarine system (MacCready & Geyer 2010).
K = 0.035 · UT · B (12.2)
where
• UT amplitude of depth-averaged tidal flow • B the length scale of an estuarine system, which is the smaller of channel width or tidal excursion. If users have tidal flow information, Equation (12.2) is a practical option to estimate spatially explicit tidal dispersion coefficient for their study area. Advective transport and tidal dispersion combine to determine physical transport of a water quality state variable. Physical transport processes take the same mathematical forms for all water quality state variables. That is, physical transport processes do not depend on the nature of the substances as long as the substances do not affect the water movement. The last term (S) in Equation (12.1) represent material-specific biogeochemical processes consisting of internal sources/sinks (SI) and external sources/sinks (SE). SI is primarily due to the kinetic processes and SE includes pol- lutant loading into and removal from a water body. Different water quality state variables are affected by different biogeochemical processes and require appropriate kinetic formulations for each of the source and sink terms (Park 1996).
12.3.2 Biogeochemical Processes
Unlike physical transport processes, each water quality state variable is determined by different biogeochemical pro- cesses and requires appropriate kinetic formulations (Park 1996). The kinetic formulations are mostly empirical and thus have to be refined with the advances in our understanding of the representing kinetic processes. The InVEST marine water quality model provides users with a flexible framework to update or add biogeochemical processes for their target materials. An example of biogeochemical processes for pathogen simulation is given below.
Pathogens
Pathogens are disease-causing microorganisms that include bacteria, viruses, and protozoa, and can originate from many sources including sewage treatment plants, urban runoff, storm sewers, failing septic systems, industrial dis- charges, and contaminated sediments (Ji 2008). Contaminated water by pathogens is responsible for the spread of many contagious diseases, and understanding the dispersal and fate of pathogens is one of main concerns for water quality management. The studies of pathogens in surface water usually focus on indicator organisms such as fecal coliforms, E. coli or enterococci, and often consider a simple decay for the biogeochemical processes (Thomann & Mueller 1987): W S = −K C + (12.3) B VOL where • C concentration of indicator organism (organism count m−3) −1 • KB decay rate (day ) • W external load of indicator organism (organism count day−1) • VOL volume of water cell (m3)
12.3. The model 193 InVEST User Guide, Release 2.5.6
As shown in Table 2, the average decay rate of total coliform bacteria is about 1.4 day−1 in freshwater (20◦C) and 48 day−1 in seawater, but the maximum decay rate can be as large as 84 day−1 under optimal environmental conditions. Table 2. Observed decay rates of indicator organisms (modified from Table 5.9 in Thomann & Mueller 1987). −1 Indictor organisms KB(day ) Note Total coliform 0.7-3.0 (avg. 1.4) Average freshwater (20◦C) 8.0-84.0 (avg. 48.0) Seawater (20◦C) (variable temperature) 0 Total or fecal 0.0-2.4 New York Harbor Salinity: 2-18 /00 (dark) 0 2.5-6.1 New York Harbor Salinity: 15 /00 (sunlight) Fecal coliform 37.0-110.0 Seawater (sunlight) 0 E-Coli 0.08-2.0 Seawater, 10-30 /00 Salmonella 0.1 - 3.0 Stormwater (20◦C), Hamilton Bay (18◦C) Mancini (1978) made an equation to estimate decay rates of indicator bacteria as a function of salinity, temperature, sunlight and sink/resuspension.
(T −20) αI0 vs KB = [0.8 + 0.006(% sea water)]1.07 + [1 − exp(−KeH)] ± (12.4) KeH H
Where • T water temperature ( ◦C) • α sunlight coefficient −2 • I0 average solar radiation (cal cm ) −1 • Ke light extinction coefficient (m ) • H average depth (m) −1 • vs sink or resuspension rate (m day )
Users may consult Table 2 to find an appropriate KB for their application. If users have enough data for the envi- ronmental conditions (water temperature, salinity, light information, etc.), Equation (12.4) may be applied to estimate KB.
12.3.3 Boundary Condition
We need to define ocean and land boundary conditions to solve Equation (12.1) numerically. The ocean boundary (i.e., open boundary) indicates the outer boundary of the modeling domain adjacent to oceans. We assume the horizontal pollutant profile (e.g. C) is advected out of the modeling domain as a “frozen pattern”. That is
∇ · Cb = ∇ · Cb0 (12.5)
where −3 • Cb is the concentration (organism count m ) on an inner boundary point b −3 • Cb0 is the concentration (organism count m ) on the outer boundary adjacent to point b Additionally, no transport of C is allowed from or into the land.
12.3. The model 194 InVEST User Guide, Release 2.5.6
12.3.4 Numerical Solution
We solve Equation (12.1) by using first and second order central difference expansions of the derivative terms and deriving an implicit Crank-Nicolson scheme. This scheme is unconditionally stable and has a truncation error of O(∆h2) where h is the discrete grid cell size.
12.4 Limitations and simplifications
1. Assumes a steady state condition (no time variation of model results): The current version assumes a steady state condition and cannot produce time varying model outputs. So, users should be aware that the model produces a distribution of a water quality state variable once it reaches an equilibrium status under defined conditions. 2. Grid size of a water cell: A finer grid size better resolves spatial differences in model outputs. However, it requires more computation and memory. Too many grid cells may cause an out of memory error. 3. The credibility of physical transport: Reliable information on physical transport processes is critical for reasonable model results. 4. No vertical transport: The governing mass balance equation of the model considers only horizontal transport of mass; it simulates vertically averaged conditions. 5. Size of the modeling domain: If the ocean boundary is too close to the pollutant loading points, inaccurate boundary values may artificially affect the model results. We therefore recommend users to prepare the grid system such that the ocean boundary is not too close to the loading locations.
12.5 Data Needs
The following are the data needs for the Marine Water Quality Model. The model is distributed with default arguments which are defaulted in the following parameters on the tool’s first run. • Workspace: The directory to hold output and intermediate results of the particular model run. After the model run is completed the output will be located in this directory. To run multiple scenarios, create a new workspace for each scenario. • Area of Interest (AOI): An ESRI Shapefile that contains a polygon indicating the target area. The output raster will align with the area of extents of this polygon. The polygon should be projected. • Land Polygon: An ESRI Shapefile that contains a polygon indicating where the landmass lies. It should be in the same projection as the AOI polygon. • Output pixel size in meters: Horizontal grid size, which determines the output resolution of the pollutant density raster. A larger number will make the output grid coarser but the model will run faster, while a finer resolution will require more computation and memory. Try making this number larger if a model run encounters an out of memory error. • Grid Cell Depth: Grid size in a vertical direction (m), which is the layer thickness of the horizontal grid system. • Source Point Centroids: An ESRI Shapefile that contains a point layer indicating the centroids of point pollu- tant sources that must have a field called Id that indicates the unique identification number for that point. This file must be in the same projection as the AOI polygon. • Source Point Loading Table: Point source loading (g day−1 or organism count day−1) at the loading points that contains at least the headers ID and WPS which correspond to the identification number in the Source Point Centroids shapefile and the loading of pollutant at that point source.
12.4. Limitations and simplifications 195 InVEST User Guide, Release 2.5.6
• Decay Coefficient (KB): Decay rate in the unit of day−1. Users may consult Table 2 or use Equation (12.4) to estimate KB. • Dispersion Coefficients ( ET ): An ESRI Shapefile that contains a point layer with a field named E_km2_day indicating the dispersion coefficient (km2day−1) at that point as referenced in Equation (12.1). This file must be in the same projection as the AOI polygon. In a future release of this model this parameter will have x and y components. • (Optional) Advection Vectors (UV as point data): An ESRI Shapefile that contains a point layer with two fields named U_m_sec_ and V_m_sec_ which correspond to the U and V components (m s−1) of the 2D advective velocity vector as referenced in Equation (12.1). This file must be in the same projection as the AOI polygon.
12.6 Running the Model
To run the marine water quality model double click invest_marine_water_quality_biophysical.exe located in the folder entitled invest-3 in the InVEST installation directory. The main interface indicates the required and optional input arguments as described in the Data Needs section above. Click the Run button to start the model. A successful run will be indicated in the window and a file explorer will open containing the results. If you encounter any errors please email the output log to [email protected].
12.7 Interpreting Results
12.7.1 Model Outputs
Each of model output files is saved in the Output and Intermediate folders that are saved within the user- specified workspace. Output folder • Output\concentration.tif: The output raster indicating the concentration of the water quality state variable (e.g. pollutant). The units of the loading determine the units of output concentration. If the loading is expressed as the number (mass) of a targeted organism, the concentration unit is the number of organisms (mass) per volume. Intermediate folder • Intermediate\in_water.tif: This is a raster file indicating the land and water points that’s used in the calculation of the domain to discritize Equation (12.1). • Intermediate\tide_e.tif: This is a raster file with the interpolated values of the Tidal Dispersion Coefficients shapefile that are used to determine the ET values for each discritized grid cell. • Intermediate\adv_u.tif and intermediate\adv_v.tif: This is a raster with the interpolated values of the Advection Vectors (UV as point data) shapefile that are used to determine the U and V components respectively.
12.8 Case example illustrating model inputs and results
Managers and stakeholders want to estimate the distribution of fecal-coliform bacteria released from floathomes (recre- ational floating cabins, usually with untreated wastes) in sheltered areas along the west coast of Vancouver Island, BC, Canada. We have explored scenarios involving different levels of treatment (removal of fecal-coliform and thus a
12.6. Running the Model 196 InVEST User Guide, Release 2.5.6 decreased loading) and different spatial arrangements of floathomes. Figs. 1 and 2 show a status quo arrangement of floathomes in Lemmens Inlet (and, in the case of Fig. 1, the surrounding area). We used an initial assumption that the loading of the untreated wastes from the floathomes of 1 million bacteria per day. In another scenario assumption, we modeled the effects of secondary treatment of waste from two floathomes (the 23rd and 24th in Fig. 2), assuming 95% removal (thus the initial loading is 50,000 bacteria per day). Model results, i.e. the distribution of fecal-coliform bacteria given the location of floathomes shown in Fig. 2 and the modeled treatment of waste described above, are shown in Fig. 3.
Figure 1. A map of Clayoquot Sound, BC, Canada showing a status quo arrangement of floathomes (red dots). The dotted box indicates Lemmens Inlet, the region of interest for potentially rearranging floathomes and/or exploring the effects of treating wastes. Background colors indicate tidal dispersion coefficients for the region, a key model input.
12.8. Case example illustrating model inputs and results 197 InVEST User Guide, Release 2.5.6
Figure 2. Enlarged map of Lemmens Inlet, showing the location of floathomes. Source point centroids are shown with red x’s and red circles indicate treated wastes (23 and 24) assuming 95% removal of bacteria.
12.8. Case example illustrating model inputs and results 198 InVEST User Guide, Release 2.5.6
Figure 3. Map of modeled concentration of fecal coliform bacteria in Lemmens Inlet. Red circles indicate treated wastes. The results are for demonstration purposes only.
12.8. Case example illustrating model inputs and results 199 InVEST User Guide, Release 2.5.6
12.9 References
Park, K. 1996. Concept of surface water quality modeling in tidal rivers and estuaries. Environ. Eng. Res. 1:1-13. MacCready, P., and W. R. Geyer 2010. Advances in estuarine physics. The Annual Review of Marine Science 2:35-58. Thomann, R. V., and J. A. Mueller. 1987. Principles of surface water quality modeling and control. Prentice-Hall, NY. Ji Z-G (2008) Hydrodynamics and water quality: Modeling rivers, lakes, and estuaries. New Jersey: John Wiley & Sons, Inc.
12.9. References 200 Part III
Terrestrial and Freshwater Models
201 CHAPTER THIRTEEN
BIODIVERSITY: HABITAT QUALITY & RARITY
13.1 Summary
Biodiversity is intimately linked to the production of environmen- tal services. Patterns in biodiversity are inherently spatial, and as such, can be estimated by analyzing maps of land use and land cover (LULC) in conjunction with threats. InVEST models habi- tat quality and rarity as proxies for biodiversity, ultimately esti- mating the extent of habitat and vegetation types across a land- scape, and their state of degradation. Habitat quality and rarity are a function of four factors: each threat’s relative impact, the relative sensitivity of each habitat type to each threat, the distance between habitats and sources of threats, and the degree to which the land is legally protected. Required inputs include a LULC map, the sensitivity of LULC types to each threat, spatial data on the distribution and intensity of each threat and the location of protected areas. The model assumes that the legal protection of land is effective and that all threats to a landscape are additive.
13.2 Introduction
A primary goal of conservation is the protection of biodiversity, including the range of genes, species, populations, habitats, and ecosystems in an area of interest. While some consider biodiversity to be an environmental service, here we treat it as an independent attribute of natural systems, with its own intrinsic value (we do NOT monetize biodiversity in this model). Natural resource managers, corporations and conservation organizations are becoming increasingly interested in understanding how and where biodiversity and environmental services align in space and how management actions affect both. Evidence from many sources builds an overwhelming picture of pervasive biodiversity decline worldwide (e.g., Vi- tousek et al. 1997; Wilcove et al 1998; Czech et. al 2000). This evidence has prompted a wide-ranging response from both governments and civil society. Through the Rio Convention on Biodiversity, 189 nations have committed themselves to preserving the biodiversity within their borders. Yet, there is scant research on the overlap between opportunities to protect biodiversity and to sustain the environmental services so critical to these countries’ economic well-being. This is precisely the type of challenge that InVEST has been designed to address.
202 InVEST User Guide, Release 2.5.6
For managers to understand the patterns of distribution and richness across a landscape, individually and in aggregate, it is necessary to map the range or occurrences of elements (e.g. species, communities, habitats). The degree to which current land use and management affects the persistence of these elements must also be assessed in order to design appropriate conservation strategies and encourage resource management that maximizes biodiversity in those areas. There are a variety of approaches to identifying priorities for conservation with various trade-offs among them. Each of these approaches focuses on different facets of biodiversity attributes and dynamics, including habitat or vegetation- based representation (i.e., a coarse filter), maximizing the number of species “covered” by a network of conserved sites for a given conservation budget (Ando et al. 1998), identifying patterns of richness and endemism (CI hotspots), and conserving ecological processes. There is also a hybrid coarse-fine filter approach which selectively includes “fine-filter” elements such as species with unique habitat requirements who may not be adequately protected using a coarse-filter approach only (TNC and WWF ecoregional planning). The InVEST Habitat Quality and Rarity model is most relevant to “coarse filter”, or habitat-based approaches. The reasons for modeling biodiversity alongside environmental services are simple and powerful. Doing so allows us to compare spatial patterns of biodiversity and environmental services, and to identify win-win areas (i.e., areas where conservation can benefit both natural systems and human economies) as well as areas where these goals are not aligned. Further, it allows us to analyze trade-offs between biodiversity and environmental services across differing scenarios of future land use change. Land use/land cover (LULC) patterns that generate greater environmental service production may not always lead to greater biodiversity conservation (Nelson et al. 2008), and modeling future options today can help identify and avoid tradeoffs.
13.3 The Model
The InVEST biodiversity model (Tier 1) combines information on LULC and threats to biodiversity to produce habitat quality and rarity maps. This approach generates two key sets of information that are useful in making an initial assessment of conservation needs: the relative extent and degradation of different types of habitat types in a region and changes across time. This approach further allows rapid assessment of the status of and change in a proxy for more detailed measures of biodiversity status. If habitat changes are taken as representative of genetic, species, or ecosystem changes, the user is assuming that areas with high quality habitat will better support all levels of biodiversity and that decreases in habitat extent and quality over time means a decline in biodiversity persistence, resilience, breadth and depth in the area of decline. The habitat rarity model indicates the extent and pattern of natural land cover types on the current or a potential future landscape vis-a-vis the extent of the same natural land cover types in some baseline period. Rarity maps allow users to create a map of the rarest habitats on the landscape relative to the baseline chosen by the user to represent the mix of habitats on the landscape that is most appropriate for the study area’s native biodiversity. The model requires basic data that are available virtually everywhere in the world, making it useful in areas for which species distribution data are poor or lacking altogether. Extensive occurrence (presence/absence) data may be available in many places for current conditions. However, modeling the change in occurrence, persistence, or vulnerability of multiple species under future conditions is often impossible or infeasible. While a habitat approach leaves out the detailed species occurrence data available for current conditions, several of its components represent advances in functionality over many existing biodiversity conservation planning tools. The most significant is the ability to characterize the sensitivity of habitats types to various threats. Not all habitats are affected by all threats in the same way, and the InVEST model accounts for this variability. Further, the model allows users to estimate the relative impact of one threat over another so that threats that are more damaging to biodiversity persistence on the landscape can be represented as such. For example, grassland could be particularly sensitive to threats generated by urban areas yet moderately sensitive to threats generated by roads. In addition, the distance over which a threat will degrade natural systems can be incorporated into the model. Model assessment of the current landscape can be used as an input to a coarse-filter assessment of current conservation needs and opportunities. Model assessment of potential LULC futures can be used to measure potential changes in habitat extent, quality, and rarity on a landscape and conservation needs and opportunities in the future.
13.3. The Model 203 InVEST User Guide, Release 2.5.6
13.3.1 How it works
Habitat quality
We define habitat as “the resources and conditions present in an area that produce occupancy – including survival and reproduction – by a given organism (Hall et al. 1997:175).” Habitat quality refers to the ability of the environment to provide conditions appropriate for individual and population persistence, and is considered a continuous variable in the model, ranging from low to medium to high, based on resources available for survival, reproduction, and population persistence, respectively (Hall et al 1997). Habitat with high quality is relatively intact and has the structure and function within the range of historic variability. Habitat quality depends on a habitat’s proximity to human land uses and the intensity of these land uses. Generally, habitat quality is degraded as the intensity of nearby land-use increases (Nelleman 2001, McKinney 2002, Forman et al. 2003). The model runs using raster data, or a gridded map of square cells. Each cell in the raster is assigned a LULC type, which can be a natural (unmanaged) cover or a managed cover. LULC types can be given at any level of classification detail. For example, grassland is a broad LULC definition that can be subdivided into pasture, restored prairie, and residential lawn types to provide much more LULC classification detail. While the user can submit up to 3 raster maps of LULC, one each for a baseline, current, and future period, at a minimum the current LULC raster map has to be submitted. The user defines which LULC types can provide habitat for the conservation objective (e.g., if forest breeding birds are the conservation objective then forests are habitat and non-forest covers are not habitat). Let Hj indicate the habitat suitability of LULC type j. Which LULC types should be considered habitat? If considering biodiversity generally or if data on specific biodiversity-habitat relationships are lacking, you can take a simple binary approach to assigning habitat to LULC types. A classic example would be to follow an island-ocean model and assume that the managed land matrix sur- rounding remnant patches of unmanaged land is unusable from the standpoint of species (e.g., MacArthur and Wilson 1967). In this case a 0 would be assigned to managed LULC types in the matrix (i.e., non-habitat) and a 1 to unman- aged types (i.e., habitat). Under this modeling scheme habitat quality scores are not a function of habitat importance, rarity, or suitability; all habitat types are treated equally. Model inputs are assumed to not be specific to any particular species or species guild, but rather apply to biodiversity generally. More recent research suggests that the matrix of managed land that surrounds patches of unmanaged land can signifi- cantly influence the “effective isolation” of habitat patches, rendering them more or less isolated than simple distance or classic models would indicate (Ricketts 2001, Prugh et al. 2008). Modification of the matrix may provide opportu- nities for reducing patch isolation and thus the extinction risk of populations in fragmented landscapes (Franklin and Lindenmayer 2009). To model this, a relative habitat suitability score can be assigned to a LULC type ranging from 0 to 1 where 1 indicates the highest habitat suitability. A ranking of less than 1 indicates habitat where a species or functional group may have lower survivability. Applying this second approach greatly expands the definition of habitat from the simple and often artificial binary approach (e.g., “natural” versus “unnatural”) to include a broad spectrum of both managed and unmanaged LULC types. By using a continuum of habitat suitability across LULC types, the user can assess the importance of land use management on habitat quality holistically or consider the potential importance of “working” (or managed) landscapes. If a continuum of habitat suitability is relevant, weights with a roster of LULC on a landscape must be applied in reference to a particular species guild of group. For example, grassland songbirds may prefer a native prairie habitat above all other habitat types (the habitat score for the LULC prairie (Hprarie) equals 1), but will also make use of a managed hayfield or pasture in a pinch (the habitat score for the LULC hayfield (Hhayfield) and pasture (Hpasture) equals 0.5). However, mammals such as porcupines will find prairie unsuitable for breeding and feeding. Therefore, if specific data on species group-habitat relationships are used, the model output refers to habitat extent and quality for the species or group in the modeled set only. Besides a map of LULC and data that relates LULC to habitat suitability, the model also requires data on habitat threat density and its affects on habitat quality. In general, we consider human modified LULC types that cause habitat fragmentation, edge, and degradation in neighboring habitat threats. For example, the conversion of a habitat LULC to non-habitat LULC reduces the size and continuity of neighboring habitat patches. Edge effects refer to changes
13.3. The Model 204 InVEST User Guide, Release 2.5.6
in the biological and physical conditions that occur at a patch boundary and within adjacent patches. For example, adjacent degraded non-habitat LULC parcels impose “edge effects” on habitat parcels and can have negative impacts within habitat parcels by, for example, facilitating entry of predators, competitors, invasive species, or toxic chemicals and other pollutants. Another example: in many developing countries roads are a threat to forest habitat quality on the landscape because of the access they provide to timber and non-timber forest harvesters. Each threat source needs to be mapped on a raster grid. A grid cell value on a threat’s map can either indicate intensity of the threat within the cell (e.g., road length in a grid cell or cultivated area in a gird cell) or simply a 1 if the grid cell contains the threat in a road or crop field cover and 0 otherwise. Let ory indicate threat r’s “score” in grid cell y where r = 1, 2, ..., R indexes all modeled degradation sources. All mapped threats should be measured in the same scale and metric. For example, if one threat is measured in density per grid cell then all degradation sources should be measured in density per grid cell where density is measured with the same metric unit (e.g., km and km2). Or if one threat is measured with presence/absence (1/0) on its map then all threats should be mapped with the presence/absence scale. The impact of threats on habitat in a grid cell is mediated by four factors. 1. The first factor is the relative impact of each threat. Some threats may be more damaging to habitat, all else equal, and a relative impact score accounts for this (see Table 1 for a list of possible threats). For instance, urban areas may be considered to be twice as degrading to any nearby habitats as agricultural areas. A degradation source’s weight, wr, indicates the relative destructiveness of a degradation source to all habitats. The weight wr can take on any value from 0 to 1. For example, if urban area has a threat weight of 1 and the threat weight of roads is set equal to 0.5 then the urban area causes twice the disturbance, all else equal, to all habitat types. To reiterate, if we have assigned species group-specific habitat suitability scores to each LULC then the threats and their weights should be specific to the modeled species group. 2. The second mitigating factor is the distance between habitat and the threat source and the impact of the threat across space. In general, the impact of a threat on habitat decreases as distance from the degradation source increases, so that grid cells that are more proximate to threats will experience higher impacts. For example, assume a grid cell is 2 km from the edge of an urban area and 0.5 km from a highway. The impact of these two threat sources on habitat in the grid cell will partly depend on how quickly they decrease, or decay, over space. The user can choose either a linear or exponential distance-decay function to describe how a threat decays over space. The impact of threat r that originates in grid cell y, ry, on habitat in grid cell x is given by irxy and is represented by the following equations,
dxy irxy = 1 − if linear (13.1) dr max
2.99 irxy = exp − dxy if exponential (13.2) dr max
where dxy is the linear distance between grid cells x and y and dr max is the maximum effective distance of threat r0s reach across space. Figure 1 illustrates the relationship between the distance-decay rate for a threat based on the maximum effective distance of the threat (linear and exponential). For example, if the user selects an exponential decline and the maximum impact distance of a threat is set at 1 km, the impact of the threat on a grid cell’s habitat will decline by ~ 50% when the grid cell is 200 m from r’s source. If irxy > 0 then grid cell x is in degradation source ry’s disturbance zone. (If the expontential funcion is used to describe the impact of degradation source r on the landscape then the model ignores values of irxy that are very close to 0 in order to expedite the modeling process.) To reiterate, if we have assigned species group-specific habitat suitability scores to each LULC then threat impact over spece should be specific to the modeled species group. 3. The third landscape factor that may mitigate the impact of threats on habitat is the level of legal / institutional / social / physical protection from disturbance in each cell. Is the grid cell in a formal protected area? Or is it
13.3. The Model 205 InVEST User Guide, Release 2.5.6
Figure 13.1: Figure 1. An example of the relationship between the distance-decay rate of a threat and the maximum effective distance of a threat.
inaccessible to people due to high elevations? Or is the grid cell open to harvest and other forms of disturbance? The model assumes that the more legal / institutional / social / physical protection from degradation a cell has, the less it will be affected by nearby threats, no matter the type of threat. Let βx ∈ [0, 1] indicate the level of accessibility in grid cell x where 1 indicates complete accessibility. As decreases the impact that all threats will have in grid cell x decreases linearly. It is important to note that while legal / institutional / social / physical protections often do diminish the impact of extractive activities in habitat such as hunting or fishing, it is unlikely to protect against other sources of degradation such as air or water pollution, habitat fragmentation, or edge effects. If the threats considered are not mitigated by legal / institutional / social / physical properties then you should ignore this input or set βx = 1 for all grid cells x. To reiterate, if we have assigned species group-specific habitat suitability scores to each LULC then the threats mitigation weights should be specific to the modeled species group. 4. The relative sensitivity of each habitat type to each threat on the landscape is the final factor used when gen- erating the total degradation in a cell with habitat (in Kareiva et al. 2010 habitat sensitivity is referred to by its inverse, “resistance”). Let Sjr ∈ [0, 1] indicate the sensitivity of LULC (habitat type) j to threat r where values closer to 1 indicate greater sensitivity. The model assumes that the more sensitive a habitat type is to a threat, the more degraded the habitat type will be by that threat. A habitat’s sensitivity to threats should be based on general principles from landscape ecology for conserving biodiversity (e.g., Forman 1995; Noss 1997; Lindenmayer et al 2008). To reiterate, if we have assigned species group-specific habitat suitability scores to each LULC then habitat sensitivity to threats should be specific to the modeled species group.
Therefore, the total threat level in grid cell x with LULC or habitat type j is given by Dxj,
R Yr ! X X wr D = r i β S (13.3) xj PR y rxy x jr r=1 y=1 r=1 wr
0 0 where y indexes all grid cells on r s raster map and Yr indicates the set of grid cells on r s raster map. Note that each threat map can have a unique number of grid cells due to variation in raster resolution If Sjr = 0 then Dxj is not a function of threat r. Also note that threat weights are normalized so that the sum across all threats weights equals 1.
By normalizing weights such that they sum to 1 we can think of Dxj as the weighted average of all threat levels in grid cell x. The map of Dxj will change as the set of weights we use change. Please note that two sets of weights will only differ if the relative differences between the weights in each set differ. For example, set of weights of 0.1, 0.1, and 0.4 are the same as the set of weights 0.2, 0.2, and 0.8. A grid cell’s degradation score is translated into a habitat quality value using a half saturation function where the user must determine the half-saturation value. As a grid cell’s degradation score increases its habitat quality decreases. Let
13.3. The Model 206 InVEST User Guide, Release 2.5.6
Figure 13.2: Table 1. Possible degradation sources based on the causes of endangerment for American species classi- fied as threatened or endangered by the US Fish and Wildlife Service. Adapted from Czech et al. 2000.
13.3. The Model 207 InVEST User Guide, Release 2.5.6
the quality of habitat in parcel x that is in LULC j be given by Qxj where,
z !! Dxj Qxj = Hj 1 − z z (13.4) Dxj + k
and z (we hard code z = 2.5) and k are scaling parameters (or constants). Qxj is equal to 0 if Hj = 0. Qxj increases in Hj and decreases in Dxj. Qxj can never be greater than 1. The k constant is the half-saturation constant and is z Dxj set by the user. The parameter k is equal to the D value where 1 − z z = 0.5 . For example, if k = 5 then Dxj +k z Dxj 1− z z = 0.5 when Dxj = 5. In the biodiversity model interface we set k = 0.5 but the user can change it (see Dxj +k note in Data Needs section, #8). If you are doing scenario analyses, whatever value you chose for k the first landscape you run the model on, that same k must be used for all alternative scenarios on the same landscape. Similarly, whatever spatial resolution you chose the first time you run the model on a landscape use the same value for all additional model runs on the same landscape. If you want to change your choice of k or the spatial resolution for any model run then you have to change the parameters for all model runs, if you are comparing multiple scenarios on the same landscape.
Habitat Rarity
While mapping habitat quality can help to identify areas where biodiversity is likely to be most intact or imperiled, it is also critical to evaluate the relative rarity of habitats on the landscape regardless of quality. In many conservation plans, habitats that are rarer are given higher priority, simply because options and opportunities for conserving them are limited and if all such habitats are lost, so too are the species and processes associated with them. The relative rarity of a LULC type on a current or projected landscape is evaluated vis-a-vis a baseline LULC pattern. A rare LULC type on a current or projected map that is also rare on some ideal or reference state on the landscape (the baseline) is not likely to be in critical danger of disappearance, whereas a rare LULC type on a current or projected map that was abundant in the past (baseline) is at risk. In the first step of the rarity calculation we take the ratio between the current or projected and past (baseline) extents of each LULC type j. Subtracting this ratio from one, the model derives an index that represents the rarity of that LULC class on the landscape of interest.
Nj Rj = 1 − (13.5) Njbaseline
where Nj is the number of grid cells of LULC j on the current or projected map and Njbaseline gives the number of grid cells of LULC j on the baseline landscape. The calculation of Rj requires that the baseline, current, and/or projected LULC maps are all in the same resolution. In this scoring system, the closer to 1 a LULC’s R score is, the greater the likelihood that the preservation of that LULC type on the current or future landscape is important to biodiversity conservation. If LULC j did not appear on the baseline landscape then we set Rj = 0.
Once we have a Rj measure for each LULC type, we can quantify the overall rarity of habitat type in grid cell x with:
X X Rx = σxjRj (13.6) x=1
where σxj = 1 if grid cell x is in LULC j on a current or projected landscape and equals 0 otherwise.
13.3. The Model 208 InVEST User Guide, Release 2.5.6
13.3.2 Limitations and simplifications
In this model all threats on the landscape are additive, although there is evidence that, in some cases, the collective impact of multiple threats is much greater than the sum of individual threat levels would suggest. Because the chosen landscape of interest is typically nested within a larger landscape, it is important to recognize that a landscape has an artificial boundary where the habitat threats immediately outside of the study boundary have been clipped and ignored. Consequently, threat intensity will always be less on the edges of a given landscape. There are two ways to avoid this problem. One, you can choose a landscape for modeling purposes whose spatial extent is significantly beyond the boundaries of your landscape of interest. Then, after results have been generated, you can extract the results just for the interior landscape of interest. Or the user can limit themselves to landscapes where degradation sources are concentrated in the middle of the landscape.
13.4 Data needs
The model uses seven types of input data (five are required). 1. Current LULC map (required). A GIS raster dataset, with a numeric LULC code for each cell. The LULC raster should include the area of interest, as well as a buffer of the width of the greatest maximum threat distance. Otherwise, locations near the edge of the area of interest may have inflated habitat quality scores, because threats outside the area of interested are not properly accounted for. The dataset should be in a projection where the units are in meters and the projection used should be defined. Name: it can be named anything. Format: standard GIS raster file (e.g., ESRI GRID or IMG), with LULC class code for each cell (e.g., 1 for forest, 2 for agriculture, 3 for grassland, etc.). The LULC class codes should be in the grid’s ‘value’ column. The raster should not contain any other data. The LULC codes must match the codes in the “Sensitivity of land cover types to each threat” table below (input # 7). Sample Data Set: \InVEST\Biodiversity\Input\lc_samp_cur_b 2. Future LULC map (optional): A GIS raster dataset that represents a future projection of LULC in the land- scape. This file should be formatted exactly like the “current LULC map” (input #1). LULC that appears on the current and future maps should have the same LULC code. LULC types unique to the future map should have codes not used in the current LULC map. Again, the LULC raster should include the area of interest, as well as a buffer of the width of the greatest maximum threat distance. Otherwise, locations near the edge of the area of interest may have inflated habitat quality scores, because threats outside the area of interested are not properly accounted for. Name: it can be named anything. Format: standard GIS raster file (e.g., ESRI GRID or IMG), with LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.). The LULC class codes should be in the raster’s ‘value’ column. Sample data set: \InVEST\Biodiversity\Input\lc_samp_fut_b 3. Baseline LULC map (optional): A GIS raster dataset of LULC types on some baseline landscape with a numeric LULC code for each cell. This file should be formatted exactly like the “current LULC map” (input #1). The LULCs that are common to the current or future and baseline landscapes should have the same LULC code across all maps. LULC types unique to the baseline map should have codes not used in the current or future LULC map. Again, the LULC raster should include the area of interest, as well as a buffer of the width of the greatest maximum threat distance. Otherwise, locations near the edge of the area of interest may have inflated habitat quality scores, because threats outside the area of interested are not properly accounted for. If possible the baseline map should refer to a time when intensive mamagement of the land was relatively rare. For example, a map of LULC in 1851 in the Willamette Valley of Oregon, USA, captures the LULC pattern on the
13.4. Data needs 209 InVEST User Guide, Release 2.5.6 landscape before it was severely modified to for massive agricultural production. Granted this landscape had been modified by American Indian land clearing practices such as controlled fires. Name: it can be named anything. Format: standard GIS raster file (e.g., ESRI GRID or IMG), with LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.). The LULC class codes should be in the grid ‘value’ column. Sample data set: \InVEST\Biodiversity\Input\lc_samp_bse_b 4. Threat data (required): A table of all threats you want the model to consider. The table contains information on the each threat’s relative importance or weight and its impact across space. Name: file can be named anything File Type: *.dbf or *.xls if using ArcGIS 9.3 Rows: each row is a degradation source Columns: each column contains a different attribute of each degradation source, and must be named as follows: 1. THREAT: the name of the specific threat. Threat names must not exceed 8 characters. 2. MAX_DIST: the maximum distance over which each threat affects habitat quality (measured in km). The impact of each degradation source will decline to zero at this maximum distance. 3. WEIGHT: the impact of each threat on habitat quality, relative to other threats. Weights can range from 1 at the highest, to 0 at the lowest. Sample Data Set: \Invest\Biodiversity\Input\threats_samp.dbf Example: Hypothetical study with three threats. Agriculture degrades habitat over a larger distance than roads do, and has a greater overall magnitude of impact. Further, paved roads attract more traffic than dirt roads and thus are more destructive to nearby habitat than dirt roads. THREAT MAX_DIST WEIGHT dirt_rd 2 0.1 Paved_rd 4 0.4 Agric 8 1 5. Sources of threats(s) (required): GIS raster file of the distribution and intensity of each individual threat. You will have as many of these maps as you have threats. These thresat maps should cover the area of interest, as well as a buffer of the width of the greatest maximum threat distance. Otherwise, locations near the edge of the area of interest may have inflated habitat quality scores, because threats outside the area of interested are not properly accounted for. Each cell in the raster contains a value that indicates the density or presence of a threat within it (e.g., area of agriculture, length of roads, or simply a 1 if the grid cell is a road or crop field and 0 otherwise). All threats should be measured in the same scale and units (i.e., all measured in density terms or all measured in presence/absence terms and not some combination of metrics). The extent and resolution of these raster datasets does not need to be identical to that of the scenario maps (the LULCs map from inputs #1, #2, or #3). In cases where the threats and LULC map resolutions vary, the model will use the resolution and extent of the LULC cover map. InVEST will not prompt you for these rasters in the tool interface. It will instead automatically find and use each one, based on names in the “Threats data” table (input # 4). Therefore, these threat maps need to be in a file named “input” that is one level below the workspace identified in the model interface (see below). Please do not leave any area on the threat maps as ‘No Data’. If an area has not threat set the area’s threat level equal to 0. If you are analyzing habitat quality for more than one LULC scenario (e.g., a current and future map or a baseline, current, and future map) then you need a set of threat layers for each modeled scenario. Add a “c” at the end of the raster for all “current” threat layers, a “f” for all future threat layers, and a “b” for all “baseline” threat layers. If you
13.4. Data needs 210 InVEST User Guide, Release 2.5.6 do not use such endings then the model assumes the degradation source layers correspond to the current map. If a threat noted in the Threats data table (input # 4) is inappropriate for the LULC scenario that you are analyzing (e.g., industrial development on a Willamette Valley pre-settlement map from 1851) then enter a threat map for that time period that has all 0 values. If you do not include threat maps for a submitted LULC scenario then the model will not calculate habitat quality on the scenario LULC map. Finally, note that we assume that the relative weights of threats and sensitivity of habitat to threats do not change over time (we only submit one Threat data table and one Habitat sensitivity data table (inputs # 4 and # 7)). If you want to change these over time then you will have to run the model multiple times. Name: the name of each raster file should exactly match the name of a degradation source in the rows of the Threats data table (input #2) above with the added “_b” (baseline), “_c” (current), or “_f” (future) to indicate the threat map’s period. File name cannot be longer than 7 characters if using a GRID format. Format: standard GIS raster file (e.g., ESRI GRID or IMG), with a relative degradation source value for each cell from that particular degradation source. The “Value” column indicates the relative degradation source that cell shows. File location: files must be saved in a folder titled “input” within the model’s workspace (see below). Sample data sets: \Invest\Biodiversity\Input\crp_c; crp_f; rr_c; rr_f; urb_c; urb_f; rot_c; rot_f; prds_c; prds_f; srds_c; srds_f; lrds_c; lrds_f. By using these sets of inputs we are running a habitat quality and rarity analysis for the current and future LULC scenario maps. A habitat quality map will not be generated for the baseline map because we have not submitted any threat layers for the baseline map. The name ‘crp’ refers to cropland, ‘rr’ to rural residential, ‘urb’ to urban, ‘rot’ to rotation forestry, ‘prds’ to primary roads, ‘srds’ to secondary roads, and ‘lrds’ to light roads. 6. Accessibility to sources of degradation (optional): A GIS polygon shapefile containing data on the relative protection that legal / institutional / social / physical barriers provide against threats. Polygons with minimum accessibility (e.g., strict nature reserves, well protected private lands) are assigned some number less than 1, while polygons with maximum accessibility (e.g., extractive reserves) are assigned a value 1. These polygons can be land management units or a regular array or hexagons or grid squares. Any cells not covered by a polygon will be assumed to be fully accessible and assigned values of 1. File type: GIS polygon shapefile. Name: file can be named anything. Rows: each row is a specific polygon on the landscape Columns: 1. ID: unique identifying code for each polygon. FID also works. 2. ACCESS: values between 0 and 1 for each parcel, as described above. Sample data set: \InVEST\Biodiversity\Input\access_samp.shp 7. Habitat types and sensitivity of habitat types to each threat (required). A table of LULC types, whether or not they are considered habitat, and, for LULC types that are habitat, their specific sensitivity to each threat. Name: file can be named anything File type: *.dbf or *.xls if using ArcMAP 9.3 Rows: each row is a LULC type. Columns: columns contain data on land use types and their sensitivities to threatss. Columns must be named according to the naming conventions below. 1. LULC: numeric code for each LULC type. Values must match the codes used in the LULC maps submitted in inputs # 1 through 3. All LULC types that appear in the current, future, or baseline maps (inputs # 1 through 3) need to appear as a row in this table.
13.4. Data needs 211 InVEST User Guide, Release 2.5.6
2. NAME: the name of each LULC 3. HABITAT: Each LULC is assigned a habitat score, Hj, from 0 to 1. If you want to simply classify each LULC as habitat or not without reference to any particular species group then use 0s and 1s where a 1 indicates habitat. Otherwise, if sufficient information is available on a species group’s habitat preferences, assign LULC a relative habitat suitability score from 0 to 1 where 1 indicates the highest habitat suitability. For example a grassland songbird may prefer a native prairie habitat above all other habitat types (prairie is given a “Habitat” score of 1 for grassland birds), but will also use a managed hayfield or pasture in a pinch (managed hayfield and pasture is given a “Habitat” score of 0.5 for grassland birds). 4. L_THREAT1, L_THREAT2, etc.: The relative sensitivity of each habitat type to each threat. You will have as many columns named like this as you have threat, and the italicized portions of names must match row names in the “Threat data” table noted above (input # 4). Values range from 0 to 1, where 1 represents high sensitivity to a threat and 0 represents no sensitivity. Note: Even if the LULC is not considered habitat, do not leave its sensitivity to each threat as Null or blank, instead enter a 0 and the model will convert it to NoData. Sample data set: \Invest\Biodiversity\Input\sensitivity_samp.dbf Example: A hypothetical study with four LULC and three threats. In this example we treat woodlands and forests as (absolute) habitat and bare soil and cultivated areas as (absolute) non-habitat. Forest mosaic is the most sensitive (least resistant) habitat type, and is more sensitive to dirt roads than paved roads or agriculture (0.9 versus 0.5 and 0.8). We enter 0’s across all threats for the two developed land covers, base soil and cultivation. LULC NAME HABITAT L_AG L_ROAD L_DIRT_RD 1 Bare Soil 0 0 0 0 2 Closed Woodland 1 0.5 0.2 0.4 3 Cultivation 0 0 0 0 4 Forest Mosaic 1 0.8 0.8 0.5 8. Half-saturation constant (required): This is the value of the parameter k in equation (4). By default it is set to 0.5 but can be set equal to any positive number. In general, you want to set k to half of the highest grid cell degradation value on the landscape. To perform this model calibration you will have to the run the model once to find the highest degradation value and set k for your landscape. For example, if a preliminary run of the model generates a degradation map where the highest grid-cell degradation level is 1 then setting k at 0.5 will produce habitat quality maps with the greatest variation on the 0 to 1 scale (this helps with visual representation of heterogeneity in quality across the landscape). It is important to note that the rank order of grid cells on the habitat quality metric is invariant to your choice of k. The choice of k only determines the spread and central tendency of habitat quality scores. Please make sure to use the same value of k for all runs that involve the same landscape. If you want to change your choice of k for any model run then you have to change the parameters for all model runs.
13.5 Running the Model
Before running the Biodiversity Model, first make sure that the InVEST toolbox has been added to your ARCMAP document, as described in the Getting Started chapter of this manual. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. Specifically, you will need (1) a current LULC raster file showing the location of different LULC types in the landscape; (2) a future LULC raster if you wish to project future habitat quality and rarity across the landscape; (3) a baseline LULC map if you wish to express habitat rarity on the current and future landscapes or measure habitat extent and quality on the baseline landscape; (4) a threat data table denoting the intensity and distance over which a degradation source occurs; (5) grids showing the spatial distribution of each threat on each submitted map (current, future, and baseline); (6) a shapefile indicating the
13.5. Running the Model 212 InVEST User Guide, Release 2.5.6 relatively accessibility to an area based on protection; (7) a table indicating the habitat suitability for each LULC and the sensitivity of each habitat type to each threat; and (8) a numeric value indicating the half-saturation constant. • Create a workspace: You must create a folder in your workspace called “input” and place all your input files here, including all your threat maps. If this is your first time using InVEST and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you unzipped the InVEST files to your C-drive (as described in the Getting Started chapter), you should see a folder called /Invest/biodiversity. This folder should be your workspace. The input files are in a folder called /Invest/biodiversity/input and in /Invest/base_data. • Open an ARCMAP document to run your model. • Find the INVEST toolbox in ARCTOOLBOX. ARCTOOLBOX should be open in ARCMAP, but if it is not, click on the ARCTOOLBOX symbol. See the Getting Started chapter if you do not see the InVEST . • Click once on the plus sign on the left side of the INVEST toolbox to see the list of tools expand. Double-click on Biodiversity.
• An interface will pop up like the one above that indicates default file names, but you can use the file buttons to browse to your data. When you place your cursor in each space, you can read a description of the data requirements in the right side of the interface. In addition, refer to the Data Needs section above for information on data formats.
13.5. Running the Model 213 InVEST User Guide, Release 2.5.6
• Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires you to enter some data. • After entering all values as required, click on OK. The script will run, and its progress will be indicated by a “Progress dialogue.” • Upon successful completion of the model, you will see new folders in your workspace called “intermediate” and “output.” These folders contain several raster grids which are described in the next section. • Load the output grids into ARCMAP using the ADD DATA button. • You can change the SYMBOLOGY of a layer by right-clicking on the layer name in the table of contents, selecting PROPERTIES, and then SYMBOLOGY. There are many options here to change the file’s appearance. • You can also view the attribute data of output files by right clicking on a layer and selecting OPEN ATTRIBUTE TABLE.
13.6 Interpreting Results
13.6.1 Final Results
Final results are found in the Output folder within the Workspace specified for this module. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix.
13.6. Interpreting Results 214 InVEST User Guide, Release 2.5.6
• degrad_cur – Relative level of habitat degradation on the current landscape. A high score in a grid cell means habitat degradation in the cell is high relative to other cells. Grid cells with non-habitat land cover (LULC with Hj = 0) get a degradation score of 0. This is a mapping of degradation scores calculated with equation (3). • qual_cur – Habitat quality on the current landscape. Higher numbers indicate better habitat quality vis-a-vis the distribution of habitat quality across the rest of the landscape. Areas on the landscape that are not habitat get a quality score of 0. This quality score is unitless and does not refer to any particular biodiversity measure. This is a mapping of habitat qulaity scores calculated with equation (4). • rarity_cur – Relative habitat rarity on the current landscape vis-a-vis the baseline map. This output is only created if a baseline LULC map is submitted (input # 3). This map gives each grid cell’s value of Rx (see equation (6)). The rarer the habitat type in a grid cell is vis-a-vis its abundance on the baseline landscape, the higher the grid cell’s rarity_cur value.
Optional Output Files
If you are running a future scenario (i.e., you have provided input # 2 and future LULC scenario threat layers), you will also see degrad_fut and qual_fut in the output folder as well. Further, if you have submitted a baseline LULC map (input # 3) as well, you will also see the raster rarity_fut in the output folder. If you have entered a baseline map (input # 3) and threat layers for the baseline (input # 4)), then you will find the rasters degrad_bse AND qual_bse in the output folder. Recall, if you are setting Hj for all LULC j on a continuum between 0 and 1 based on the habitat suitability for a particular species group then these results are only applicable to that species group.
Modifying output and creating a landscape biodiversity score
The model output doesn’t provide landscape-level quality and rarity scores for comparing the baseline, current, and future LULC scenarios. Instead the user must summarize habitat extent and quality and rarity scores for each land- scape. At the simplest level, a habitat quality landscape score for a LULC scenario is simply the aggregate of all grid cell-level scores under the scenario. In other words, we can sum all grid-level quality scores on the qual_bse (if available), qual_cur, and qual_fut (if available) maps and then compare scores. A map may have a higher aggregate quality score for several reasons. For one, it may just have more habitat area. However, if the amount of habitat across any two scenarios is approximately the same then a higher landscape quality score is indicative of better overall quality habitat. Scores for certain areas on a landscape could also be compared. For example, we could compare aggregate habitat quality scores in areas of the landscape that are known to be in the geographic ranges of species of interest. For example, suppose we have geographic range maps of 9 species and have submitted current and future LULC scenario maps to the Tier 1 biodiversity model. In this case we would determine 18 aggregate habitat quality scores, once for each modeled species under each scenario. Let Gscur indicate the set of grid cells on the current landscape that are in s‘ range. Then the average habitat quality score in species s‘ range on the current landscape is given by, PGscur x=1 Qxjcur Qs = (13.7) cur Gscur
where Qxjcur indicates the habitat quality score on parcel x in LULC j on the current landscape and Qxjcur =0 if qual_cur for x is “No Data”. The average range-normalized habitat quality score for all 9 species on the current landscape would be given by,
X X Rx = σxjRj (13.8) x=1
13.6. Interpreting Results 215 InVEST User Guide, Release 2.5.6
Then we would repeat for the future landscape with the grid cells in set Gs_fut for each species s and the set of Qxjfut .
13.7 Biodiversity 3.0 Beta
We are working on the next generation of the InVEST platform and the biodiversity model exists in this form. You can try out the 3.0 version of Biodiversity by navigating to your Windows Start Menu -> All Programs -> InVEST -> Terrestrial -> Biodiversity. The interface does not require ArcGIS and the results can be explored with any GIS tool including ArcGIS, QuantumGIS, and others. In an earlier version of InVEST this tool had a decay parameter to differentiate between linear and exponential decay. That parameter has been removed in this version of InVEST and the biodiversity model exclusively uses exponential decay.
13.8 References
Ando, A, J. Camm, S. Polasky, and A. Solow. 1998. Species distributions, land values, and efficient conservation. Science 279:2126-2128. Czech, B., P. R. Krausman, and P. K. Devers. 2000. Economic Associations among Causes of Species Endangerment in the United States. Bioscience 50:593-601. Forman, R. 1995. Land Mosaics: The Ecology of landscapes and regions. Cambridge Univ Press. New York. Forman, R. 2003. Road ecology: science and solutions. Island Press. New York, New York. Franklin, J.F. and D. B. Lindenmayer. 2009. Importance of matrix habitats in maintaining biological diversity. Pro- ceedings of the National Academy of Sciences 106:349-350. Hall, L.S., Krausman, P.R. and Morrison, M.L. 1997. The habitat concept and a plea for standard terminology. Wildlife Society Bulletin 25(1):173-182. Lindenmayer, D., Hobbs, R., Montague-Drake, R., Alexandra, J., Bennett, A., Burgman, M., Cae, P., Calhoun, A., Cramer, V., Cullen, P. 2008. A checklist for ecological management of landscapes for conservation. Ecology Letters 11:78-91. MacArthur, R., E. 0. Wilson. 1967. The theory of island biogeography. Princeton University Press, Princeton, NJ. Mckinney, M.L. 2002. Urbanization, biodiversity, and conservation. BioScience 52:883-890. Nelleman C, Kullered L, Vistnes I, Forbes B, Foresman T, Husby E, Kofinas G, Kaltenborn B, Rouaud J, Magomedova M, Bobiwash R, Lambrechts C, Schei P, Tveitdal S, Gron O, Larsen T. 2001. GLOBIO. Global methodology for mapping human impacts on the biosphere. UNEP/DEWA/TR.01-3. Nelson, E., S. Polasky, D. J. Lewis, A. J. Plantinga, E. Lonsdorf, D. White, D. Bael & J. J. Lawler. 2008. Efficiency of incentives to jointly increase carbon sequestration and species conservation on a landscape. Proc. Nat. Acad. Sci. 105: 9471-9476. Noss, R. F., M. A. Connell, and D. D. Murphy. 1997. The science of conservation planning: habitat conservation under the endangered species act. Island Press. Prugh, L., K. Hodges, A. Sinclair, and J. Brashares. 2008. Effect of habitat area and isolation on fragmented animal populations. Proceedings of the National Academy of Sciences 105:20770. Ricketts, T. H. 2001. The Matrix Matters: Effective Isolation in Fragmented Landscapes. American Naturalist 158:87- 99. Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human Domination of Earth’s Ecosystems. Science 277:494.
13.7. Biodiversity 3.0 Beta 216 InVEST User Guide, Release 2.5.6
Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying Threats to Imperiled Species in the United States. Bioscience 48:607-615.
13.8. References 217 CHAPTER FOURTEEN
CARBON STORAGE AND SEQUESTRATION
14.1 Summary
Terrestrial ecosystems, which store more carbon than the atmo- sphere, are vital to influencing carbon dioxide-driven climate change. The InVEST model uses maps of land use and land cover types and data on wood harvest rates, harvested product degrada- tion rates, and stocks in four carbon pools (aboveground biomass, belowground biomass, soil, dead organic matter) to estimate the amount of carbon currently stored in a landscape or the amount of carbon sequestered over time. Additional data on the market or social value of sequestered carbon and its annual rate of change, and a discount rate can be used in an optional model that estimates the value of this environmental service to society. Limitations of the model include an oversimplified carbon cycle, an assumed linear change in carbon sequestration over time, and potentially inaccurate discounting rates.
14.2 Introduction
Ecosystems regulate Earth’s climate by adding and removing greenhouse gases (GHG) such as CO2 from the atmosphere. In fact, forests, grasslands, peat swamps, and other terrestrial ecosystems collectively store much more carbon than does the atmosphere (Lal 2002). By storing this carbon in wood, other biomass, and soil, ecosystems keep CO2 out of the atmosphere, where it would contribute to climate change. Beyond just storing carbon, many systems also continue to accumulate it in plants and soil over time, thereby “sequestering” additional carbon each year. Disturbing these systems with fire, disease, or vegetation conversion (e.g., land use / land cover (LULC) conversion) can release large amounts of CO2. Other management changes, like forest restoration or alternative agricultural practices, can lead to the storage of large amounts of CO2. Therefore, the ways in which we manage terrestrial ecosystems are critical to regulating our climate. As with all other models for which InVEST provides estimates of value, we are focused on the social value of carbon sequestration and storage. Terrestrial-based carbon sequestration and storage is perhaps the most widely recognized of all environmental services (Stern 2007, IPCC 2006, Canadell and Raupach 2008, Capoor and Ambrosi 2008, Hamilton et al. 2008, Pagiola 2008). The social value of a sequestered ton of carbon is equal to the social damage avoided by not releasing the ton of carbon into the atmosphere (Tol 2005, Stern 2007). Calculations of social cost are complicated
218 InVEST User Guide, Release 2.5.6 and controversial (see Weitzman 2007 and Nordhaus 2007b), but have resulted in value estimates that range from USD $9.55 to $84.55 per metric ton of CO2 released into the atmosphere (Nordhaus 2007a and Stern 2007, respectively). In addition to the social value of carbon sequestration and storage, there are several emerging markets for carbon based on both regulation and voluntary demand. The Kyoto Protocol – the current treaty addressing international climate change – includes a mechanism for establishing projects that sequester carbon to earn credits, which they then can sell to others needing to offset their own CO2 emissions. As a result of the Kyoto Protocol, the European Union Emissions Trading Scheme (EU ETS) emerged to allow the regulated firms of the EU to trade their emissions allowances. The Chicago Climate Exchange (CCX) emerged in the United States, which is not a signatory party of the Kyoto Protocol. The CCX allows interested parties to trade emissions offsets that have been certified on a voluntary basis. The EU ETS and the CCX hadprices of around 25 Euros and $USD 6 per metric ton of CO2, respectively as of April 2008. In addition to these centralized markets, there is a substantial over-the-counter market for voluntary carbon offsets. For details about the price of these offsets, see Conte and Kotchen (2010). Currently these markets only apply to carbon sequestration (i.e., the additional storage of carbon over time), but there is increased interest in financial incentives to avoid release of carbon from ecosystems in the first place, so- called “reduced emissions from deforestation and degradation” or “REDD” (Gibbs et al. 2007, Mollicone et al. 2007, Mackey et al. 2008). This option was accepted during the last meeting of the parties to the UN Framework Convention on Climate Change and is likely to be written in to the follow up agreement to the Kyoto Protocol. Payments for REDD would financially reward forest owners for reversing their planned deforesting and thinning actions (Sedjo and Sohngen 2007, Sohngen et al. 2008). Issues of accounting and verification have slowed the emergence of REDD markets, but many are anticipating them with private transactions.
While market prices are one way to estimate the value of CO2 sequestration, these prices will reflect policies, subsidies, and other factors, and therefore will only indicate the true value of this service to society by chance (Murray et al. 2007). For this reason, we recommend that users rely on the avoided damages associated with the emission of CO2 into the atmosphere rather than prices in existing carbon markets to estimate the social value of carbon sequestration and storage. Managing landscapes for carbon storage and sequestration requires information about how much and where carbon is stored, how much carbon is sequestered or lost over time, and how shifts in land use affect the amount of carbon stored and sequestered over time. Since land managers must choose among sites for protection, harvest, or development, maps of carbon storage and sequestration are ideal for supporting decisions influencing these environmental services. Such maps can support a range of decisions by governments, NGOs, and businesses. For example, governments can use them to identify opportunities to earn credits for reduced (carbon) emissions from deforestation and degradation (REDD). Knowing which parts of a landscape store the most carbon would help governments efficiently target in- centives to landowners in exchange for forest conservation. Additionally, a conservation NGO may wish to invest in areas where high levels of biodiversity and carbon sequestration overlap (Nelson et al. 2008). A timber company may also want to maximize its returns from both timber production and REDD carbon credits (Plantinga and Birdsey 1994), in which case they could use the InVEST timber production model in tandem with the carbon model to assess management options. Figure 1. Conceptual model of carbon storage and sequestration. Parameters depicted in color are included in the InVEST model, while those in gray are not.
14.3 The Model
Carbon storage on a land parcel largely depends on the sizes of four carbon “pools:” aboveground biomass, be- lowground biomass, soil, and dead organic matter (Fig. 1). The InVEST Carbon Storage and Sequestration model aggregates the amount of carbon stored in these pools according to the land use maps and classifications produced by the user. Aboveground biomass comprises all living plant material above the soil (e.g., bark, trunks, branches, leaves). Belowground biomass encompasses the living root systems of aboveground biomass.Soil organic matter is the organic component of soil, and represents the largest terrestrial carbon pool. Dead organic matter includes litter as well as lying and standing dead wood. A fifth optional pool included in the model applies to parcels that produce harvested
14.3. The Model 219 InVEST User Guide, Release 2.5.6
wood products (HWPs) such as firewood or charcoal or more long-lived products such as house timbers or furniture. Tracking carbon in this pool is useful because it represents the amount of carbon kept from the atmosphere by a given product. Using maps of land use and land cover types and the amount of carbon stored in carbon pools, this model estimates: the net amount of carbon stored in a land parcel over time; the total biomass removed from a harvested area of the parcel, and the market and social values of the carbon sequestered in remaining stock. Limitations of the model include an oversimplified carbon cycle, an assumed linear change in carbon sequestration over time, and potentially inaccurate discounting rates. Biophysical conditions important for carbon sequestration such as photosynthesis rates and the presence of active soil organisms are also not included in the model (Fig. 1).
14.3.1 How it works
The model runs on a gridded map of cells called raster format in GIS. If the HWP pool is included in the analysis, a polygon map of harvest parcels is also modeled. Each cell in the raster is assigned a land use and land use and land cover (LULC) type such as forest, pasture, or agricultural land. Each harvest polygon is assigned harvest type referring to the harvested product, harvest frequency, and product decay rates. After running the model in raster format, results can be summarized to practical land units such as individual properties, political units, or watersheds. For each LULC type, the model requires an estimate of the amount of carbon in at least one of the four fundamental pools described above. If the user has data for more than one pool, the modeled results will be more complete. The model simply applies these estimates to the LULC map to produce a map of carbon storage in the carbon pools included. For the fifth carbon pool, HWP, model values are defined for each parcel (polygon) and not for each LULC. For each parcel the user indicates the amount of biomass, in terms of carbon, removed per harvest, the frequency of harvests, and the rate at which the products that contain carbon degrade. With these data, the model calculates the amount of stored carbon that originated in a parcel but now resides in finished products such as houses or furniture. The model converts parcel level HWP carbon values into a grid cell layer that spatially matches the grid system used for the other four carbon storage pools. The model aggregates the carbon in each of the five pools, providing an estimate of total carbon storage in each grid cell and across the whole landscape. If carbon storage data for a given pool are not mapped, then total carbon storage
14.3. The Model 220 InVEST User Guide, Release 2.5.6 will be underestimated. The model also outputs the total biomass and volume of wood removed from each harvested parcel up to the year associated with the modeled landscape. If the user provides both a current and future LULC map, then the net change in carbon storage over time (sequestration and loss) and its social value can be calculated. To estimate this change in carbon sequestration over time, the model is simply applied to the current landscape and a projected future landscape, and the difference in storage is calculated, map unit by map unit. If multiple future scenarios are available, the differences between the current and each alternate future landscape can be compared. Outputs of the model are expressed as Mg of carbon per grid cell, or if desired, the value of sequestration in dollars per grid cell. We strongly recommend using the social value of carbon sequestration if the user is interested in expressing sequestration in monetary units. The social value of a sequestered ton of carbon is the social damage avoided by not releasing the ton of carbon into the atmosphere. The market value may be applicable if the user is interested in identifying the value of the landscape for trading under current market conditions. The market value of terrestrial- based carbon sequestration is the price per metric ton of carbon traded in marketplaces such as the Chicago Climate Exchange (ECX). The valuation model estimates the economic value of sequestration (not storage) as a function of the amount of carbon sequestered, the monetary value of each unit of carbon, a monetary discount rate, and the change in the value of carbon sequestration over time (Fig. 1). Thus, valuation can only be done in the carbon model if you have a future scenario. Valuation is applied to sequestration, not storage, because current market prices relate only to carbon sequestration. Discount rates are multipliers that typically reduce the value of carbon sequestration over time. The first type of discounting, the standard economic procedure of financial discounting, reflects the fact that people typically value immediate benefits more than future benefits due to impatience and uncertain economic growth. The second discount rate adjusts the social value of carbon sequestration over time. This value will change as the impact of carbon emissions on expected climate change-related damages changes. If we expect carbon sequestered today to have a greater impact on climate change mitigation than carbon sequestered in the future this second discount rate should be positive. On the other hand, if we expect carbon sequestered today to have less of an impact on climate change mitigation than carbon sequestered in the future this second discount rate should be negative.
14.3.2 Uncertainty analysis
In many cases, limited data can make it difficult to determine precisely the amount of carbon in different pools. To accomodate such data limitations, the model optionally performs uncertainty analysis. If users choose to run the model with uncertainty analysis, then inputs and outputs are both affected. Input data when using uncertainty analysis must specify probability distributions for amount of carbon in different pools. For each carbon pool type, input data must specify both the mean estimate, which represents the expected carbon amount, and the standard deviation, which represents the uncertainty for the estimate. When running uncertainty analysis, model outputs include all of the original outputs of the non-uncertainty model, such as a raster mapping total carbon per grid cell. To calculate the outputs produced by the non-uncertainty model, the uncertainty model uses the user-provided mean estimates for the carbon pools and ignores the standard deviation data. In addition to these outputs, which use only the mean estimate data, the uncertainty model also produces two types of uncertainty outputs: (1) ‘confidence’ rasters to indicate areas where we are confident that sequestration or emissions will occur, and (2) standard deviations for outputs.
Confidence raster
When provided with uncertainty data, the carbon model will produce a ‘confidence’ output raster, which uses both the mean and the standard deviation data and highlights areas where it is highly likely that storage will either increase or decrease. The model uses a user-provided confidence threshold as the minimum probability for which grid cells should be highlighted.
14.3. The Model 221 InVEST User Guide, Release 2.5.6
To compute the probability that storage increases or decreases in a particular grid cell, we use the LULC data and the HWP data (if present) to construct probability distributions for the current carbon storage in the grid cell and the future carbon storage in the cell. The current carbon storage is distributed with mean µcurr and standard deviation σcurr. The future carbon storage is distributed with mean µfut and standard deviation σfut. Since we assume that both are normally distributed, we can compute the probability p that future carbon storage is greater than current carbon storage as follows:
µ − µ p = Φ( fut cur ) q 2 2 σcurr + σfut
where Φ is the cumulative distribution function of the normal distribution. This value of p for a particular grid cell is then used to determine how confident we are that storage will either increase or decrease in that cell.
Output standard deviations
In addition to the confidence maps, the uncertainty model will also compute standard deviations for output quantities such as carbon storage, carbon sequestration, and value of sequestered carbon. These standard deviations are computed via a Monte Carlo simulation. For each iteration of the simulation, the model samples a value for carbon per grid cell for each LULC type, given the input normal distribution for that LULC type. Then, for the given iteration of the simulation, the model will assign that amount of carbon to each pixel with the given LULC type, and compute the amount of carbon stored in each scenario. In other words, for a given run of the iteration, all pixels with the same LULC type will be assigned the same amount of carbon; that amount will be chosen by taking a random sample from the input normal distribution. Outputs for each run of the simulation are recorded, and then analyzed to extract data about mean and standard deviation, which are reported in the output summary file. This feature is not supported by the current model if HWP analysis is enabled.
14.3.3 REDD scenario analysis
The carbon model can optionally perform scenario analysis according to a framework of Reducing Emissions from Forest Degradation and Deforestation (REDD) or REDD+. REDD is a scheme for emissions reductions under which countries that reduce emissions from deforestation can be financially compensated. REDD+ builds on the original REDD framework by also incorporating conservation, sustainable forest management, and enhancement of existing carbon stocks. To perform REDD scenario analysis, the model requires three LULC maps: one for the current scenario, one for a future baseline scenario, and one for a future scenario under a REDD policy. The future baseline scenario is used to compute a reference level of emissions against which the REDD scenario can be compared. Depending on the specifics on the desired REDD framework, the baseline scenario can be generated in a number of different ways; for instance, it can be based on historical rates of deforestation or on projections. The REDD policy scenario map reflects future LULC under a REDD policy to prevent deforestation and enhance carbon sequestration. Based on these three LULC maps for current, baseline, and REDD policy scenarios, the carbon biophysical model produces a number of outputs. First, it produce rasters for total carbon storage for each of the three LULC maps. Sec- ond, it produces two sequestration rasters. One sequestration raster indicates sequestration from the current scenario to the baseline scenario. The other sequestration raster indicates sequestration from the current scenario to the REDD policy scenario.
14.3. The Model 222 InVEST User Guide, Release 2.5.6
If uncertainty analysis is enabled, the carbon biophysical model will also produce two additional confidence rasters. One raster represents regions where the model is confident (beyond the user-provided confidence threshold) that carbon storage will either increase or decrease in the transition from the current scenario to the future baseline scenario. The second raster represents regions where the model is confident that carbon storage will either increase or decrease in the transition from the current scenario to the REDD policy scenario. The model currently does not support REDD scenario analysis together with harvested wood product analysis. There- fore, if REDD scenario analysis is enabled, HWP analysis will be disabled.
14.3.4 Limitations and simplifications
The model greatly oversimplifies the carbon cycle which allows it to run with relatively little information, but also leads to important limitations. For example, the model assumes that none of the LULC types in the landscape are gaining or losing carbon over time. Instead it is assumed that all LULC types are at some fixed storage level equal to the average of measured storage levels within that LULC type. Under this assumption, the only changes in carbon storage over time are due to changes from one LULC type to another or from the harvest of wood products. Therefore, any grid cell that does not change its LULC type and is at a wood harvest steady-state will have a sequestration value of 0 over time. In reality, many areas are recovering from past land use or are undergoing natural succession. The problem can be addressed by dividing LULC types into age classes (essentially adding more LULC types), such as three ages of forest. Then, parcels can move from one age class to the other in scenarios and change their carbon storage values as a result. A second limitation is that because the model relies on carbon storage estimates for each LULC type, the results are only as detailed and reliable as the LULC classification used. Carbon storage clearly differs among LULC types (e.g., tropical forest vs. open woodland), but often there can also be significant variation within a LULC type. For example, carbon storage within a “tropical moist forest” is affected by temperature, elevation, rainfall, and the number of years since a major disturbance (e.g., clear-cut or forest fire). The variety of carbon storage values within coarsely defined LULC types can be partly recovered by using a LULC classification system and related carbon pool table which stratifies coarsely defined LULC types with relevant environmental and management variables. For example, forest LULC types can be stratified by elevation, climate bands or time intervals since a major disturbance. Of course, this more detailed approach requires data describing the amount of carbon stored in each of the carbon pools for each of the finer LULC classes. Another limitation of the model is that it does not capture carbon that moves from one pool to another. For example, if trees in a forest die due to disease, much of the carbon stored in aboveground biomass becomes carbon stored in other (dead) organic material. Also, when trees are harvested from a forest, branches, stems, bark, etc. are left as slash on the ground. The model assumes that the carbon in wood slash “instantly” enters the atmosphere. With respect to its estimates of carbon in HWPs, the model is constrained by the fact that users may assign only one harvest rate (e.g., 50 Mg of wood per harvest where a harvest occurs every 2 years) and only one decay rate (e.g., the wood harvested from the parcel over the years is always used to make the same product that decays at the same rate) to each parcel. In reality, harvested parcels will exhibit variation in harvest and decay rates over time. The model also does not account for the greenhouse gasses (GHGs) emitted from the transportation of harvested wood from its initial harvest site to its final destination, the conversion of raw wood into finished products, or agriculture-related activities such as from tractors and livestock. Annual GHG emissions from agricultural land use can be calculated with the InVEST Agriculture Production Model, due to be released soon. The uncertainty model has a few limitations. First, it assumes that the probability distribution for amount of carbon in different pools is normally distributed. This may not be the case; for instance, predictions for carbon amounts may be asymmetric distributions. If this is the case, users should choose a normal distribution that best approximates the desired distribution. In addition, uncertainty in input data is currently limited to carbon pools. The model does not yet handle uncertainty in LULC maps or HWP data. The carbon model also assumes that every carbon pool across different LULC types is independent. If, in reality, estimates for carbon pools are consistently too high or too low, then error may be greater than predicted by the model. Finally, while most sequestration follows a nonlinear path such that carbon is sequestered at a higher rate in the first
14.3. The Model 223 InVEST User Guide, Release 2.5.6 few years and a lower rate in subsequent years, the model’s economic valuation of carbon sequestration assumes a linear change in carbon storage over time. The assumption of a constant rate of change will tend to undervalue the carbon sequestered, as a nonlinear path of carbon sequestration is more socially valuable due to discounting than a linear path (Fig.2).
Figure 2: The model assumes a linear change in carbon storage (the solid line), while the actual path to the year T’s carbon storage level may be non-linear (like the dotted line). In this case t can indicate the year of the current landscape and T the year of the future landscape. With positive discounting, the value of the modeled path (the solid line) is less valuable than the actual path. Therefore, if sequestration paths tend to follow the dotted line, the modeled valuation of carbon sequestration will underestimate the actual value of the carbon sequestered.
14.4 Data needs
This section outlines the map and data tables required by the model, including the economic data that the tool interface will prompt the user to enter. See Appendix for detailed information on data sources and pre-processing. 1. Current land use/land cover (LULC) map (required): A GIS raster dataset, with a LULC code for each cell. The dataset should be projected in meters and the projection used should be defined. Name: file can be named anything, but avoid spaces Format: standard GIS raster file (e.g., ESRI GRID or IMG), with LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.) These codes must match LULC codes in the tables below. LULC class codes should be in the ‘LULC’ column of the dataset. Sample data set: \Invest\Base_Data\Terrestrial\lulc_samp_cur The model requires the following two pieces of information about the LULC map which are prompted for in the interface. • The year depicted by the LULC map, for use in calculating sequestration and economic values (labeled “Year of current land cover” in the interface). • The spatial resolution (desired cell size in meters) at which you would like the model to run (labeled “Resolution (optional)”). You can only define a new resolution that is coarser than the resolution of the LULC map (this is the default resolution).
14.4. Data needs 224 InVEST User Guide, Release 2.5.6
2. Carbon pools (required): A table of LULC classes, containing data on carbon stored in each of the four fundamental pools for each LULC class. Carbon storage data can be collected from field estimates from local plot studies, extracted from meta-analyses on specific habitat types or regions, or found in general published tables (e.g., IPCC, see Appendix). If information on some carbon pools is not available, pools can be estimated from other pools, or omitted by leaving all values for the pool equal to 0. If a forest is regularly harvested for woody biomass, the estimates of carbon biomass in the aboveground, belowground, and dead organic matter pools should reflect this fact. For example, suppose one of the LULC types is a plantation forest that tends to have one-tenth of its area clear-cut every year. The above- ground and belowground estimates of carbon biomass for this LULC type should reflect the fact that only 9/10ths of the area occupied by plantation forests will be covered by trees at any point in time. For notes on calculating standard deviation for the uncertainty model, see the Appendix for data sources for carbon stocks. Name: file can be named anything File type: *.csv or *.dbf Rows: each row is a LULC class Columns: each column contains a different attribute of each LULC class, and must be named as follows: • lucode: code of land use/land cover class (e.g., 1 for forest, 3 for grassland, etc.). The LULC code should match the LULC codes from the current LULC map (dataset #1 above) • LULC_name: descriptive name of LULC class (optional) To run the model without uncertainty analysis, the following columns are required: • C_above: amount of carbon stored in aboveground biomass (in Mg ha-1) • C_below: amount of carbon stored in belowground biomass (in Mg ha-1) • C_soil: amount of carbon stored in soil (in Mg ha-1) • C_dead: amount of carbon stored in dead organic matter (in Mg ha-1) To run the model with uncertainty analysis, the following columns are required: • C_above_mean: estimated amount of carbon stored in aboveground biomass (in Mg ha-1) • C_above_sd: standard deviation to measure uncertainty in the amount of carbon in aboveground biomass (in Mg ha-1) • C_below_mean: estimated amount of carbon stored in belowground biomass (in Mg ha-1) • C_below_sd: standard deviation to measure uncertainty in the amount of carbon in belowground biomass (in Mg ha-1) • C_soil_mean: estimated amount of carbon stored in soil (in Mg ha-1) • C_soil_sd: standard deviation to measure uncertainty in the amount of carbon in soil (in Mg ha-1) • C_dead_mean: estimated amount of carbon stored in dead organic matter (in Mg ha-1) • C_dead_sd: standard deviation to measure uncertainty in the amount of carbon in dead organic matter (in Mg ha-1) Note: The unit for all carbon pools is Mg of elemental carbon ha-1. This means that if your data source -1 has information on Mg of CO2 stored ha , you need to convert those numbers to elemental carbon by -1 multiplying Mg of CO2 stored ha by 0.2727. Sample data set (without uncertainty): \Invest\Carbon\Input\carbon_pools_samp.csv Sample data set (with uncertainty): \Invest\Carbon\Input\carbon_pools_samp_uncertain.csv
14.4. Data needs 225 InVEST User Guide, Release 2.5.6
Example (without uncertainty): Hypothetical study with five LULC classes. Class 1 (Forest) contains the most carbon in all pools. In this example, carbon stored in above- and below-ground biomass differs strongly among land use classes, but carbon stored in soil varies less dramatically. lucode LULC_name C_above C_below C_soil C_dead 1 Forest 140 70 35 12 2 Coffee 65 40 25 6 3 Pasture/grass 15 35 30 4 4 Shrub/undergrowth 30 30 30 13 5 Open/urban 5 5 15 2 Example (with uncertainty): As above, but with standard deviations to measure uncertainty in carbon pool estimates. lu- LULC_name C_above_meanC_above_sdC_below_meanC_below_sdC_soil_meanC_soil_sdC_dead_meanC_dead_sd code 1 Forest 140 20 70 10 35 5 12 2 2 Coffee 65 5 40 10 25 5 6 2 3 Pas- 15 3 35 5 30 5 4 1 ture/grass 4 Shrub/undergrowth30 5 30 7 30 8 13 3 5 Open/urban 5 1 5 1 15 2 2 0.5 Confidence threshold (for uncertainty): The uncertainty model also requires an additional confidence threshold parameter which is input directly through the tool interface rather than through a file. This is used as the minimum probability of storage increase or decrease for which we highlight cells in the ‘conf’ output file. 3. Current harvest rates map (optional). A GIS shape file of polygons (parcels in our vernacular), contains data on: 1. Parcel ID 2. Amount of carbon, in the form of woody biomass, typically removed from the parcel over the course of a harvest period 3. Date that the modeler wants to begin accounting for wood harvests in the parcel 4. Frequency of harvest periods in the parcel in the past 5. Average decay rate of products made from the wood harvested from a parcel 6. Average carbon density of the wood removed form the parcel in the past 7. Average tree volume per ton of wood removed form the parcel in the past. The GIS polygon map should only delineate parcels that have been harvested; all other portions of the landscape should be ignored. Note that unlike the current LULC map,this file contains multiple data for each individual harvest parcel on the landscape. The amount of carbon that is removed, on average, during each harvest period can be estimated from plot surveys, market demand analyses, community surveys, or based on expert opinion. Decay rates can be estimated from literature reports (see sources in Appendix) or also based on expert opinion if necessary. If multiple types of wood products are harvested from a polygon, the user should average the rates of decay or focus on the product with the slowest decay rate (since that will affect storage the most). Because only woody biomass is included in the harvest portion of the model, it is not necessary to include harvest or decay rates for herbaceous products. If you are unable or uninterested in estimating carbon stored in harvested wood products, you do not need to supply this table and the model will ignore this pool. Name: file can be named anything File type: GIS polygon shapefile
14.4. Data needs 226 InVEST User Guide, Release 2.5.6
Rows: each row is a specific polygon on the landscape. Columns: columns contain attributes related to harvested wood products and must be named as follows: 1. FID: unique identifying code for each polygon (parcels in our vernacular). 2. Cut_cur: The amount of carbon typically removed from a parcel during a harvest period (measured in Mg ha-1; the model will sum across the area of each parcel). This amount should only include the portion of the wood’s carbon that is removed from the parcel (e.g., the carbon in the wood delivered to a saw mill). In other words, the slash and other waste from a wood harvest should be ignored because the model assumes that its carbon content is lost to the atmosphere instantly (the “cur” at the end of this attribute is used to relate it to the “current” LULC map). 3. Start_date: The first year the carbon removed from a forest will be accounted for in the HWP pool. The first year should coincide with a year in which wood was actually harvested from the parcel. If wood was harvested from a parcel in 1995, 2000, and 2005 and the LULC map being evaluated is from 2005 then St_date can equal 1995, 2000, or 2005; it is your choice. 4. Freq_cur: The frequency, in years, with which the Cut_cur amount is harvested. If the value is 1 then the Cut_cur amount is removed annually from the parcel, if 5 then every 5 years, etc. 5. Decay_cur: The half-life of wood products harvested, measured in years. 6. C_den_cur: The carbon density in the harvested wood (MgC Mg-1of dry wood). Typically, the statistic ranges between 0.43 and 0.55 (see table 4.3 of IPCC (2006)). If C_den_cur is not known for a parcel set it equal to 0.5. 7. BCEF_cur: An expansion factor that translates the mass of harvested wood into volume of harvested wood (Biomass Conversion Expansion Factor). The expansion factor is measured in Mgof dry wood per m3 of wood and is a function of stand type and stand age. If you do not have data on this expansion factor you can use the BCEFR row in table 4.5 of IPCC (2006). Otherwise, set this expansion factor equal to 1 for each parcel. Sample data set: \Invest\Carbon\Input\harv_samp_cur.shp Example: A hypothetical study of carbon storage in HWP for four forest parcels that have experienced harvests in the past. Assume the current LULC map we are using corresponds to the year 2005. Parcels 1, 2, and 3 are forests that are managed for timber production. Each managed forest experiences a cut every 5th year where Cut_cur gives the amount of carbon (Mg ha-1) in the portion of the wood that is removed every fifth year. The fourth parcel is a source of firewood and wood is cut from the parcel continuously. Thus, for this parcel we estimate the annual rate of carbon removed from the forest for firewood. For the first three parcels, we began to account for carbon removal in 1995. For the final parcel we began accounting for HWP in 2000. (Recall that the calculation of HWP_cur, Bio_HWP_cur, and Vol_HWP_cur does not include the 2005 harvest; that carbon is still on the land.) FID Cut_cur Start_date Freq_cur Decay_cur C_den_cur BCEF_cur 1 75 1995 5 30 0.5 1 2 50 1995 5 35 0.5 1 3 50 1995 5 50 0.5 1 4 45 2000 1 1 0.5 1 We measure the carbon stored in HWP that originated from parcel x on the current landscape with the following equation:
yr_cur−start_date ru( F req_cur )−1 X x HWP _curx = Cut_curx × f(Decay_curx; yr_cur − start_datex − (t × F req_curx)) t=0 (14.1)
14.4. Data needs 227 InVEST User Guide, Release 2.5.6
where HWP _curx is measured in Mg ha-1, yr_cur is short for “Year of current land cover”, t indexes the number of harvest periods, and ru indicates that any fraction should be rounded up to the next integer value. The function
1 − e−ωx f(•) = (14.2) [yr_cur−start_date −(t×F req_cur )]×ω ωx × e x x x
where ωx = (loge 2/Decay_curx), measures how much of the carbon was typically removed from a par- cel (Cut_curx) during a harvest period, that occurred some number of years ago (yr_cur −start_datex − (t × F req_curx)), still remains trapped in HWP as of the current year (yr_cur) and given the current decay rate (Decay_curx). The following are several examples to show how equation (1) works. In the first instance, assume start_datex = 1983, yr_cur = 2000, and F req_curx = 4. In this case, ru yr_cur−start_date = F req_curx 17 ru 4 = ru(4.25) = 5. According to the summation term in equation (1), this means we sum over 5 harvest periods (t = 0,1,2,3,4). Given this series of t, we evaluate f at 17, 13, 9, 5, and 1 years since a harvest (we use to convert the series of t‘s into years since harvest). Alternatively, if start_datex = 1980, yr_cur = 2000, and F req_curx = 2 then ru yr_cur−start_date = ru(10) = 10. Therefore, according to equation (1), harvests that contained F req_curx Cut_curx of carbon ha-1 occurred on the parcel 20, 18, 16, 14, 12, 10, 8, 6, 4, and 2 years before the year 2000 (note that we do not include a harvest that is scheduled to occur in the current year in the HWP carbon pool; this carbon is still in situ in the current year). We use C_den_cur and BCEF _cur to measure the mass (Bio_HWP _cur) and volume (V ol_HWP _cur) of wood that has been removed from a parcel from the start_date to the current year. Bio_HWP _curfor parcel x is measured in Mg (dry matter) ha-1 and is given by:
yr_cur − start_date 1 Bio_HWP _curx = Cut_curx × ru × (14.3) F req_curx C_den_curx
and V ol_HWP _cur for parcel x is measured in m3 of wood ha-1 and is given by,
1 V ol_HWP _curx = Bio_HWP _curx × (14.4) V ol_exp_curx
As mentioned before, the model places all parcel-level values into a grid cell map that comports with the four pool storage map. 4. Future Scenarios (optional – required for valuation): If you have a LULC map (data input #1) for a future landscape scenario, then expected sequestration rates in the four major carbon pools on the landscape can be measured. Similarly, sequestration rates in the HWP carbon pool can be measured with a harvest rate map (data input #3) for this future landscape. If REDD scenario analysis is enabled, then this should represent the landscape for the future baseline scenario, against which the REDD scenario will be compared. A future land cover map (a raster dataset) should be formatted according to the same specifications as the current land cover map (input #1). If you provide a future harvest rate map then the HWP carbon pool can be tracked over time. The future harvest rate map should be formatted according to the same specifications as the current harvest rate map:
14.4. Data needs 228 InVEST User Guide, Release 2.5.6
a polygon map where values for FID, Cut_fut, Freq_fut, Decay_fut, C_den_fut, and BCEF_fut are at- yr_cur+yr_fut tributed to each parcel that is expected be harvested at some point between the year given by 2 and yr_fut where yr_fut indicates the year associated with the future land cover map (e.g., if yr_cur is 2000 yr_cur+yr_fut and fut_yr is 2050 then 2 = 2025). This means that current harvest rate map conditions hold on the landscape until the year halfway between the current and future years. The harvest variables for the yr_cur+yr_fut future will be applied in the year 2 . Note that any fraction is round down (e.g., if yr_cur is yr_cur+yr_fut 2000 and fut_yr is 2053 then 2 = 2026). The future harvest rate map does not have to retain any spatial semblance to the current harvest rate map. Nor do parcels that are harvested on the current and future maps have to have a common FID. Sample data files for future scenarios are future land cover: (\In- VEST\Base_Data\Terrestrial\lulc_samp_fut) and future harvest rate map (\In- VEST\Carbon\Input\harv_samp_fut.shp). Example: A hypothetical study of future carbon storage in HWP for four forest parcels. Continuing with current harvest rate map (2005) described above, assume the future LULC map corresponds to the year 2035. Three of the four forest parcels that have wood removed on the current landscape keep their boundaries in the future and continue to have wood removed into the future (parcels with FID 1, 3, and 4 on the current harvest rate map). However the first parcel changes its management with newCut and Freq values (Cut_curx 6= Cut_futx and F req_curx 6= F req_futx). We assume these new management yr_cur+yr_fut conditions begin in the year 2020 (given by 2 ). Parcel 2 is not expected to be harvested at yr_cur+yr_fut any point between 2 and yr_fut. Therefore, the model assumes that the harvest activity given in current harvest rate map for parcel 2 ends in 2020. In addition, the future harvest rate map includes a new harvested parcel (given by FID = 5). We assume that harvest begins there in 2020 as well. In parcels 3 and 4 harvest management does not change across the current and future landscapes. (Note that we retained the FID values across the two maps here; this is not necessary, as the ArcGIS program will perform the necessary spatial matches). FID Cut_fut Freq_fut Decay_fut C_den_fut BCEF_fut 1 50 10 30 0.5 1 3 50 5 50 0.5 1 4 45 1 1 0.5 1 5 25 2 15 0.5 1 Below we describe exactly how the futur e harvest values are calculated. If a parcel was harvested on the current landscape and is expected to be harvested on the future landscape (i.e., at some point between yr_cur+yr_fut 2 and yrf ut) then the remaining HWP carbon due to harvest from parcel x in the future year is given by:
yr_fut+yr_cur −1 −start_date ru 2 x P F req_curx HWP _futx = Cut_curx t=0 f(Decay_curx, yr_fut − start_datex − (t × F req_curx))+ yr_fut+yr_cur −1 yr_fut− ru 2 P F req_futx yr_fut+yr_cur Cut_futx t=0 f Decay_futx, yr_fut − 2 − (t × F req_futx) (14.5)
where the function f is as before. Recall that if (yr_cur + yr_fut) / 2 results in a fraction it is rounded down. Also note that equation (5) does not include a harvest that is scheduled to occur in the future year; this harvest’s carbon isin situ in this accounting. Parcels that were harvested on the current landscape but are not expected to be harvested on the future landscape may still have HWP carbon in the future year.
14.4. Data needs 229 InVEST User Guide, Release 2.5.6
The remaining HWP carbon in yr_fut on such parcels is given by the first term of equation (5):
yr_fut+yr_cur −1 −start_date ru 2 x F req_curx X HWP _futx = Cut_curx × f(Decay_curx, yr_fut − start_datex − (t × F req_curx)) t=0 (14.6)
In contrast, parcels that were not harvested on the current landscape, but are expected to be harvested on the future landscape, will have the following amount of carbon in the form of HWP in yr_fut:
yr_fut+yr_cur −1 yr_fut− ru 2 F req_futx X yr_fut + yr_cur HWP _fut = Cut_fut f Decay_fut , yr_fut − − (t × F req_fut ) x x x 2 x t=0 (14.7)
Note that this is the second term of equation (5). If a parcel was harvested on the current landscape and is expected to be harvested on the future landscape, the mass of harvested wood that has been removed from a parcel from Start_date to yr_fut is given by:
yr_fut+yrcur 2 −start_datex 1 Bio_HWP _futx = Cut_curx × ru × + F req_curx C_den_curx yr_fut+yr_cur (14.8) yr_fut− 2 1 Cut_futx × ru × F req_futx C_den_fut
However, for parcels that were harvested on the current landscape, but are not expected to be harvested on the future landscape, the mass of wood removed from a parcel from Start_date to yr_fut is given by the first term of equation (8):
yr_fut+yrcur ! ! 2 − start_datex 1 Bio_HWP _futx = Cut_curx × ru × (14.9) F req_curx C_den_curx
For parcels that were not harvested on the current landscape but are expected to be harvested on the future landscape, the mass of wood removed from a parcel from Start_date toyr_futis given by second term of equation (8):
yr_fut+yr_cur ! ! yr_fut − 2 1 Bio_HWP _futx = Cut_futx × ru × (14.10) F req_futx C_den_fut
Finally, the volume of the of wood that has been removed from a parcel from Start_date to yr_fut is given by:
yr_fut+yr_cur 2 −start_datex 1 1 V ol_HWP _futx = Cut_curx × ru × × + F req_curx C_den_curx BCEF _curx yr_fut+yr_cur yr_fut− 2 1 1 Cut_futx × ru × × F req_futx C_den_futx BCEF _futx (14.11)
14.4. Data needs 230 InVEST User Guide, Release 2.5.6
yr_fut+yr_cur ! ! 2 − start_datex 1 1 V ol_HWP _futx = Cut_curx × ru × × F req_curx C_den_curx BCEF _curx (14.12)
or
yr_fut+yr_cur ! ! yr_fut − 2 1 1 V ol_HWP _futx = Cut_futx × ru × × F req_futx C_den_futx BCEF _futx (14.13)
depending on the combination of current and future harvests (see above). We recommend that the modeler use Bio_HWP_cur and Bio_HWP_fut to refine the current and future LULC maps. Specifically, if Bio_HWP_cur or Bio_HWP_fut on a portion of the landscape are significant, then the modeler should assess whether the LULC types associated with that portion of the current or future landscape accurately reflect the biomass remaining on the landscape. For example, if the current LULC type on a portion of the landscape that has been heavily harvested in the immediate past is “closed conifer” it may be more appropriate to reclassify it as “thinned conifer” or “open conifer” on the LULC map. 5. REDD scenario LULC map (optional). REDD scenario analysis requires a LULC map for a land- scape scenario under a REDD policy. This should be formatted according to the same specifications as the current and the baseline future land cover map. The REDD scenario LULC map must be for the same year as the baseline future scenario LULC map. 6. Economic data (optional – required for valuation). Three numbers are not supplied in a table, but instead are input directly through the tool interface. (a) The value of a sequestered ton of carbon (V in the equation below), in dollars per metric ton of elemental carbon (not CO2, which is heavier, so be careful to get units right! If the social value of CO2e is $Y per metric ton, then the social value of C is $(3.67*Y) per metric ton (Labeled “Price of carbon per metric ton (optional)” in the tool interface.) For applications interested in estimating the total value of carbon sequestration, we recommend value estimates based of damage costs associated with the release of an additional ton of carbon (the social cost of carbon (SCC). Stern (2007), Tol (2009), and Nordhaus (2007a) present estimates of SCC. For example, two SCC estimates we have used from Tol (2009) are $66 and $130 (in 2010 US dollars) (Polasky et al. 2010). For applications interested in estimating the value that could be gained by trading carbon credits in the current markets, the value can be taken from the current market prices on the Chicago or European Climate Exchanges. (b) The market discount rate (r in the equation below), which reflects society’s preference for immediate benefits over future benefits (labeled “Market discount rate (%) (optional)” in the tool interface). The default value in the interface is 7% per year, which is one of the market dis- count rates recommended by the U.S. government for cost-benefit evaluation of environmental projects. However, this rate will depend on the country and landscape being evaluated. Philo- sophical arguments have been made for using a lower discount rate when modeling climate change related dynamics, which users may consider using. If the rate is set equal to 0% then monetary values are not discounted. (c) The annual rate of change in the price of carbon (c in the equation below), which adjusts the value of sequestered carbon as the impact of emissions on expected climate change-related damages changes over time. The default value in the interface is 0% (labeled “The annual rate of change in the price of carbon (%) (optional)” in the tool interface). However, settingthis rate
14.4. Data needs 231 InVEST User Guide, Release 2.5.6
greater than 0% suggests that the societal value of carbon sequestered in the future is less than the value of carbon sequestered now. It has been widely argued that GHG emissions need to be curtailed immediately to avoid crossing a GHG atmospheric concentration threshold that would lead to a 3 degree Celsius or greater change in global average temperature by 2105.Some argue that such a temperature change would lead to major disruptions in economies across the world (Stern et al. 2006). Therefore, any mitigation in GHG emissions that occurs many years from now may have no effect on whether or not this crucial concentration threshold is passed. If this is the case, C sequestration in the far future would be relatively worthless and a carbon discount rate greater than zero is warranted. Alternatively, setting the annual rate of change less than 0% (e.g., -2%) suggests that the societal value of carbon sequestered in the future is greater than the value of carbon sequestered now (this is a separate issue than the value of money in the future, a dynamic accounted for with the market discount rate). This may be the case if the damages associated with climate change in the future accelerate as the concentration of GHGs in the atmosphere increases. The value of carbon sequestration over time is given by:
yr_fut−yr_cur−1 sequestx X 1 value_seqx = V (14.14) yr_fut − yr_cur r t c t t=0 1 + 100 1 + 100
14.5 Running the Model
Before running the Carbon Storage and Sequestration model, make sure that the INVEST toolbox has been added to your ARCMAP document, as described in the Getting Started chapter. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. Specifically, you will need (1) a land cover raster file showing the location of different land cover and land use types in the landscape; and (2) a carbon pools file which denotes the amount of aboveground, belowground, and soil carbon, and carbon from dead biomass, by land cover type. Optionally, you may also include (1) a map of harvest rates; (2) economic data on the value of carbon; and (3) future land use/land cover and harvest rate data to project future carbon scenarios. • Identify workspace If you are using your own data, you need to first create a workspace, or folder for the analysis data, on your computer hard-drive. The entire pathname to the workspace should not have any spaces. All your output files will be dumped here. For simplicity, you may wish to call the folder for your workspace “carbon” and create a folder in your workspace called “input” and place all your input files here. It’s not necessary to place input files in the workspace, but advisable so you can easily see the data you use to run your model. Or, if this is your first time using the tool and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you unzipped the InVEST files to your C-drive (as described in the Getting Started chapter), you should see a folder called /Invest/carbon. This folder will be your workspace. The input files are in a folder called /Invest/carbon/input and in /Invest/base_data. • Open anARCMAP document to run your model. • Find theINVEST toolbox in ARCTOOLBOX. ARCTOOLBOX is normally open in ARCMAP, but if it is not, click on the ARCTOOLBOX symbol. See the Getting Started chapter if you don’t see the InVEST toolbox and need instructions on how to add it. • You can run this analysis without adding data to your map view, but usually it is recommended to view your data first and familiarize yourself. Add the data for this analysis to your map using the ADD DATA button and look at each file to make sure it is formatted correctly. Save your ARCMAP file as needed. * Click once on the + sign on the left side of the INVEST toolbox to expand the list of tools. Double-click on Carbon.
14.5. Running the Model 232 InVEST User Guide, Release 2.5.6
14.5. Running the Model 233 InVEST User Guide, Release 2.5.6
Carbon tool dialog • An interface will pop up like the one above. The tool shows default file names, but you can use the file buttons to browse instead to your own data. When you place your cursor in each space, you can read a description of the data requirements in the right side of the interface. In addition, refer to the Data Needs section above for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires you to enter some data. If you choose to run the optional economic valuation, all optional inputs below the checkbox become required. • After you’ve entered all values as required, click on OK. The script will run, and its progress will be indicated by a “Progress dialogue”. • Upon successful completion of the model, you will see new folders in your workspace called “intermediate” and “output.” These folders contain several raster grids. These grids are described in the Interpreting Results section. • Load the output grids into ARCMAP using the ADD DATA button. • You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting PROPERTIES, and then SYMBOLOGY. There are many options here to change the way the file appears in the map. • You can also view the attribute data of output files by right clicking on a layer and selecting OPEN ATTRIBUTE TABLE.
14.6 Interpreting Results
14.6.1 Final results
Final results are found in the Output folder within the Workspace specified for this module. If REDD scenario analysis is enabled, then files with the suffix _base represent results for the baseline future sce- nario, and files with the suffix _redd represent results for the REDD policy scenario. Model results: • summary.html: This file presents a summary of all data computed by the model. It also includes descriptions of all other output files produced by the model, so it is a good place to begin exploring and understanding model results. Because this is an HTML file, it can be opened with any web browser. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix. • tot_C_cur: This file shows the amount of carbon currently stored in Mg in each grid cell at the chosen reso- lution. This is a sum of all of the carbon pools you have included data for (above ground, below ground, soil, dead material, and harvested wood product). The lowest value can be 0 (for example, paved areas if you don’t include the soil beneath the pavement). Examine this map to see where high and low values fall. Is this what you would expect given the current land use and land cover? If not, check your input files. • tot_C_fut: This file shows the total amount of carbon that will be stored in each parcel under your future landscape scenario. It is a sum of all the carbon pools for which you have included data. The values are in Mg per grid cell. Again, the lowest value can be 0. • sequest: This file maps the difference in carbon stored between the future landscape and the current landscape – or the carbon that is sequestered during the entire given time period (i.e. this is a rate per the total time period elapsed, yr_fut – yr_cur, not per year). The values are in Mg per grid cell. In this map some values may be negative and some positive. Positive values indicate sequestered carbon, whereas negative values indicate
14.6. Interpreting Results 234 InVEST User Guide, Release 2.5.6
carbon that was lost. Areas with large negative or positive values should have the biggest changes in LULC or harvest rates. Remember that carbon emissions due to management activities (tractors burning fuel, fertilizer additions, etc.) on a parcel are NOT included in this assessment. • conf (for uncertainty model only): This file maps areas where we are confident that emissions either increase or decrease. Grid cells where we are confident that storage will increase from the current LULC map to the future LULC map have a value of 1. Grid cells where we are confident storage will decrease have a value of -1. Grid cells where we are not confident either way have a value of 0. The confidence threshold specified by the user in the initial parameters is used as the minimum probability threshold for which we highlight a region with a 1 or -1. For example, if the user specifies a confidence threshold of 95, a grid cell will receive a value of 1 only if it is at least 95% likely that storage will increase in that particular cell. • value_seq: This file maps the economic value of carbon sequestered (between the current and the future land- scape dates, yr_cur and yr_fut). The relative differences between parcels should be similar (but not identical) to sequest, but the values are in dollars per grid cell instead of Mg per grid cell. As with sequest, values may be negative, indicating the cost of carbon emissions from LULC changes to that parcel. • *_mask files (for uncertainty model only): When provided with confidence rasters, the valuation model will produce files such as seq_mask and val_mask. These files contain the raster created by ‘masking’ the sequest and value_seq rasters, respectively, with the conf confidence raster. In other words, seq_mask is identical to sequest, except that areas where the conf raster indicates low confidence are ignored (and set to ‘no data’ values). Similarly, val_mask is identical to value_seq, except that areas where the conf raster indicates low confidence are ignored. Therefore, the *mask files contain values only in those cells where we have high confidence that carbon storage will increase or decrease.
14.6.2 Intermediate results
These files independently map each of the five carbon pools that contribute to the final results for both current and future landscapes. Examining these results can help you determine which of the carbon pools are changing the most between your current and future landscapes and can help you identify areas where your data may need correcting. The unit for each of these pool outputs is Mg per grid cell. Biomass_HWP_cur and Biomass_HWP_fut are both measured in Mg dry matter per grid cell and Vol_HWP_cur and Vol_HWP_fut are both measured in m3 of wood per grid cell. lc_res_cur and lc_res_fut give the current and future LULC maps at the resolution chosen with the model interface. Finally, Carbon_dateandtime_suffix.txt is a text file that summarizes the parameter data you chose when running the Carbon Storage and Sequestration Model. The text file’s name includes “dateandtime” which means that the data and time is stamped into the text’s file name. The text file’s name also includes a “suffix” term that you choose. • C_above_cur – the current carbon stock for the aboveground pool • C_above_fut – the carbon stock for the aboveground pool for the future scenario • C_below_cur – the current carbon stock for the belowground pool • C_below_fut – the carbon stock for the belowground pool for the future scenario • C_soil_cur – the current carbon stock in soil • C_soil_fut – the carbon stock in soil for the future scenario • C_dead_cur – the current carbon stock in dead organic matter • C_dead_fut – the carbon stock in dead organic matter for the future scenario • C_HWP_cur – carbon stored in harvested wood products for current land cover • C_HWP_fut – carbon stored in harvested wood products for future scenario. • Bio_HWP_cur – biomass of wood removed since “start_date” for current land cover • Bio_HWP_fut – biomass of wood removed since “start_date” for future land cover
14.6. Interpreting Results 235 InVEST User Guide, Release 2.5.6
• Vol_HWP_cur – volume of wood removed since “start_date” for current land cover • Vol_HWP_fut – volume of wood removed since “start_date” for future land cover • lc_res_cur – the current LULC map at the resolution chosen by the user. • lc_res_fut – the future LULC map at the resolution chosen by the user. • Carbon_dateandtime_suffix.txt – a text file that summarizes the parameter data used to run the Carbon Storage and Sequestration Model.
14.7 Appendix: data sources
This is a rough compilation of data sources and suggestions for finding, compiling, and formatting data. This section should be used for ideas and suggestions only. This section is updated as new data sources and methods become available.
14.7.1 1. Land use/land cover map
The simplest categorization of LULCs on the landscape involves delineation by land cover only (e.g., cropland, temperate conifer forest, prairie). Several global and regional land cover classifications are available (e.g., Anderson et al. 1976), and often detailed land cover classification has been done for the landscape of interest. A slightly more sophisticated LULC classification could involve breaking relevant LULC types into broad age categories (e.g., forest of age 0-10 years, 11-20, 21-40, etc.). This would allow separate estimates of carbon storage for different ages. In scenarios, parcels can move from one age class to the next, crudely capturing changes in carbon storage over time. This approach requires more information, however, in- cluding carbon storage estimates for each age class for all modeled pools of carbon. A still more detailed classification could stratify LULC types by variables known to affect carbon storage within a given LULC type (e.g., montane forest 800-1000m, montane forest 1001-1200m, etc.). Rainfall, temperature, and elevation all typically influence carbon storage and sequestration (e.g., Jenny 1980, Coomes et al. 2002, Raich et al. 2006). If data are available to estimate carbon storage at different elevations, or at different levels of rainfall, temperature or other climate variables, model results will be substantially more accurate. This will typically take a large sample of plot estimates of carbon storage.
14.7.2 2. Carbon stocks
Carbon storage data should be set equal to the average carbon storage values for each LULC class. The ideal data source for all carbon stocks is a set of local field estimates, where carbon storage for all relevant stocks has been directly measured. These can be summarized to the LULC map, including any stratifi- cation by age or other variable. If these data are not available, however, there are several general data sources that can be used. Note that several sources, including IPCC (2006), report in units of biomass, while InVEST uses mass of elemental carbon. To convert metric tons of biomass to metric tons of C, multiply by a conversion factor, which varies typically from 0.43 to 0.51. Conversion factors for different major tree types and climatic regions are listed in Table 4.3 on page 4.48 of IPCC (2006). Notes on calculating standard deviation for the uncertainty model: The standard deviation values in the carbon pool table signify uncertainties in the true value for amount of carbon in different pools. There are a variety of methods to calculate standard deviation. For instance, the standard deviation can be calculated from a confidence interval; a 95 percent confidence interval, for example, is 3.92 standard deviations wide. Therefore, we can divide the width of the 95 percent confidence interval by 3.92 to
14.7. Appendix: data sources 236 InVEST User Guide, Release 2.5.6
calculate standard deviation. For more information on uncertainty analysis, see Volume 1 Chapter 3, “Uncertainties”, in IPCC (2006).
14.7.3 2.1. Carbon stored in aboveground biomass
A good but very general source of data for carbon storage is the Intergovernmental Panel on Climate Change’s (IPCC) 2006 methodology for determining greenhouse gas inventories in the Agriculture, Forestry and Other Land Use (AFOLU) sector (http://www.ipcc-nggip.iges.or.jp/public/2006gl/pdf/4_Volume4/V4_02_Ch2_Generic.pdf, IPCC 2006). To use this set of information from the IPCC, you must know your site’s climate domain and region; use data from Table 4.1 on page 4.46 and a digital copy of the Food and Agriculture Organization of the United Nations’ (FAO) eco-region map (http://www.fao.org/geonetwork/srv/en/main.home) to figure that out. Tables 5.1 through 5.3 (p. 5.9) of IPCC (2006) give estimates for aboveground biomass in agriculture land with perennial woody biomass (e.g., fruit orchards, agroforestry, etc.). Tables 4.7, 4.8, and 4.12 give aboveground biomass estimates for natural and plantation forest types. Recently, Ruesch and Gibbs (2008) mapped the IPCC (2006) aboveground biomass carbon storage data given year 2000 land cover data. Other general sources of carbon storage estimates can be found. For example, Grace et al. (2006) estimate the average aboveground carbon storage (leaf + wood) for major savanna ecosystems around the world (Table 1). Houghton (2005) gives aboveground carbon storage for natural and plantation forest types, by continent (Tables 1 and 3). Brown et al. (1989) give aboveground biomass estimatesfor tropical broadleaf forests as a function of land-use: undisturbed, logged, nonproductive (Table 7). Region-specific sources of carbon storage data are also available. Those we’ve found include: • Latin America: Malhi et al. (2006) report aboveground biomass volumes for 227 lowland forest plots in Bolivia, Brazil, Colombia, Ecuador, French Guinea, Guyana, Panama, Peru, and Venezuela. Nascimento and Laurance (2002) estimate aboveground carbon stocks in twenty 1-ha plots of Amazonian rainforest. Tiessen et al. (1998) find aboveground carbon stocks for the Brazilian savanna types Caatingas and Cerrados. • Africa: Zhang and Justice (2001) report aboveground carbon stocks for major forest and shrub LULC types for central African countries. Tiessen et al. (1998) estimates total aboveground biomass of degraded savanna in Senegal. Makundi (2001) reports mean annual incremental growth for three forest plantation types in Tanzania. Malimbwi et al. (1994) estimates aboveground carbon stocks in the miombo woodlands of Kitungalo Forest Reserve Tanzania. Munishi and Shear (2004) report aboveground carbon stocks in the Afromontane rain forests of the Eastern Arc Mountains of Tanzania. Glenday (2006) estimates aboveground carbon stocks for 3 forest types in the Kakamega National Forest of western Kenya. • North America: Smith et al. (2006) estimate aboveground carbon stocks for all major forest types in the US. • The Carbon On Line Estimator (http://www.ncasi2.org/COLE/) is a tool for calculating carbon characteristics in U.S. forests based on USDA Forest Service Forest Inventory & Analysis and Resource Planning Assessment data. With this tool, carbon characteristics can be examined at the scale of counties. Using the variables tab, aboveground, belowground, soil, or dead wood carbon pools can be selected. • Other: Coomes et al. (2002) estimate aboveground carbon stocks for native shrubland and forest types in New Zealand. One can also calculate aboveground biomass (and therefore carbon stocks) from timber inventories, which are often done by forestry ministries on a set of plots. Use the following formula to estimate the aboveground carbon stock in a forest stand that has been inventoried for its merchantable volume, where VOB is the per-hectare volume of trees in cubic meters measured from tree stump to crown point (the merchantable portion of the tree), WD is the wood density of trees (dry biomass per unit of tree volume), BEF is the ratio of total aboveground dry biomass to dry biomass of inventoried volume, and CF is the ratio of elemental carbon to dry biomass, by mass (Brown 1997). The biomass expansion factor (BEF) accounts for C stored in all other portions of the tree aboveground (e.g., branches, bark, stems, foliage, etc; the non-merchantable portions of the tree). In most cases WD for a plot is approximated with values for dominant species. Brown (1997) provides a table of WD values for many tree species in Appendix 1 of section 3 and a method for calculating BEF (Equation 3.1.4). See ECCM (2007) for an application of this FAO method to forest inventory data from eastern Tanzania. IPCC (2006) also presents estimates of ( ) where BEF values for hardwood,
14.7. Appendix: data sources 237 InVEST User Guide, Release 2.5.6 pine, conifer, and natural forest stands by eco-region are given in Table 4.5 and WD values for many species are given in Tables 4.13 and 4.14. (Use the BCEF values in Table 4.5 that are subscripted by S.) Finally, Brown et al. (1989) give BEF for tropical broadleaf forests under three land uses: undisturbed, logged, and nonproductive. Brown (1997) attaches several caveats to the use of the above equation. First, the equation is designed for inventoried stands that are closed as opposed to open (forests with sparser canopy coverage such as oak savanna). Second, VOB estimates should be a function of all tree species found in the stand, not just the economically most valuable wood. Third, trees with diameters as low as 10 centimeters at breast height (DBH = 10) need to be included in the inventory if this aboveground biomass carbon equation is to be as accurate as possible. Brown (2002) also notes that the use of a single BEF value is a simplification of the actual biomass growth process. These caveats lead Brown (2002) to recommend the use of allometric biomass equations to estimate woody above- ground biomass if available. These equations give the estimated relationship between a stand’s distribution of different- sized trees and the stand’s aboveground biomass. Brown (1997) and Brown and Schroeder (1999) provide general aboveground biomass allometric equations for all global eco-regions and the eastern US, respectively. Cairns et al. (2000) provide aboveground biomass allometric equations for LULC types in southern Mexico. Nascimento and Lau- rance (2002) estimate Amazonian rainforest aboveground biomass using allometric curves. The use of these equations requires knowledge of the distribution of tree size in a given stand. Some researchers have made use of these equations a bit easier by first relating a stand’s distribution of different-sized trees to its age and then mapping the relationship between age and aboveground biomass (i.e., ). For example, Silver et al. (2000) have estimated aboveground biomass as a function of stand age (i.e., years since afforestation/ reforestation) or previous LULC for native forest types in tropical ecosystems. Smith et al. (2006) take the transformation of allometric equations one step further by relating age to total biomass carbon (belowground plus aboveground) directly for various US forests. When using IPCC data or other similar broad data sources, one final issue to consider is how the level of anthropogenic disturbance affects carbon stocks. The aboveground C stock of highly disturbed areas will likely be lower than the stocks of undisturbed areas. It is not clear what type of disturbance levels IPCC or other such sources assume when reporting aboveground biomass estimates. If forest disturbance is an issue in the demonstration site, LULC types should be stratified by levels of disturbance. For an example of such stratification see Table 2.5, page 14 of ECCM (2007). The effect of this disturbance on C storage in harvested wood products (HWPs) is discussed below. Finally, we generally do nottreat aboveground herbaceous material as a carbon pool (e.g., grass, flowers, non-woody crops). Our working assumption is that this material does not represent a potential source of long-term storage like woody biomass, belowground biomass, and soil. Herbaceous material in general recycles its carbon too quickly.
14.7.4 2.2. Carbon stored in belowground biomass
For LULC categories dominated by woody biomass, belowground biomass can be estimated roughly with the “root to shoot” ratio of belowground to aboveground biomass. Default estimates of the root to shoot ratio are given in Table 4.4 on p. 4.49 of IPCC (2006) by eco-region. Broad estimates of this ratio are also given in Section 3.5 of Brown (1997). Some LULC types contain little to no woody biomass but substantial belowground carbon stocks (e.g., natural grass- lands, managed grasslands, steppes, and scrub/ shrub areas). In these cases the root to shoot ratio described above does not apply. Belowground estimates for these LULC types are best estimated locally, but if local data are not available some global estimates can be used. The IPCC (2006) lists total biomass (aboveground plus belowground) and above- ground biomass for each climate zone in table 6.4 (p. 6.27). The difference between these numbers is a crude estimate of belowground biomass. . Recently, Ruesch and Gibbs (2008) mapped the IPCC (2006) aboveground biomass carbon storage data given year 2000 land cover data. Several studies have compiled estimates of belowground biomass or root-to-shoot ratios for different habitat types. Among those we found: • Grace et al. (2006) estimate the total average woody and herbaceous root biomass for major savanna ecosystems around the world (Table 1). Baer et al. (2002) and Tilman et al. (2006) estimate the C stored in the roots of
14.7. Appendix: data sources 238 InVEST User Guide, Release 2.5.6
plots restored to native C4 grasses in Nebraska and Minnesota, U.S. respectively, as a function of years since restoration (see Table 2 in Baer et al. (2002) and Figure 1D in Tilman et al. (2006)). • Cairns et al. (1997) survey root-to-shoot ratios for LULC types across the world. Munishi and Shear (2004) use a ratio of 0.22 for Afromontane forests in the Eastern Arc forests of Tanzania. Malimbwi et al. (1994) use 0.20 for miombo woodlands in the same area of Tanzania. Coomes et al. (2002) use 0.25 for shrublands in New Zealand. Gaston et al. (1998) report a root-to-shoot ratio of 1 for African grass / shrub savannas.
14.7.5 2.3. Carbon stored in soil
If local or regional soil C estimates are not available, default estimates can be looked up from IPCC (2006) for agri- cultural, pasture, and managed grasslands. Table 2.3 of IPCC (2006) contains estimates of soil carbon stocks by soil type, assuming these stocks are at equilibrium and have no active land management. For cropland and grasslandLULC types, this default estimate can be multiplied by management factors, listed in Tables 5.5 and 6.2 of IPCC (2006). For all other LULC types and their related management schemes, the IPCC (2006) assumes no management factors. There are alternative global-level sources of soil carbon data. Post et al. (1982) report carbon stocks in the first meter of soil by Holdridge Life Zone Classification System (GIS map of these Zones available at http://www.ngdc.noaa.gov/seg/cdroms/ged_iia/datasets/a06/lh.htm). Silver etal. (2000) have estimated soil carbonas a function of years since afforestation / reforestation for native forest types in tropical ecosystems. Grace et al. (2006) estimate the soil carbon for major savanna types around the world (Table 1). Detwiler (1986) lists soil carbon for tropical forest soils in Table 2. Several region-specific studies also report soil carbon stocks. Those we’ve found include: • North America: Smith et al. (2006) estimate soil C for every 5-year increment up to 125 years since afforesta- tion/ reforestation for all major forest types and forest management practices in each region of the U.S. Others include McLauchlan et al. (2006); Tilman et al. (2006); Fargione et al (2008); Schuman et al. (2002); and Lal (2002). • Africa: Houghton and Hackler (2006) give soil C for 5 LULC forest types (Rain Forest; Moist Forest Dry; Forest; Shrubland; and Montane Forest) in sub-Saharan Africa that have retained their natural cover and for forest areas that have been converted to croplands, shifting cultivation, and pasture. Vagen et al. (2005) provides soil C estimates for various LULC types in sub-Saharan Africa. • South America: Bernoux et al. (2002) estimated soil C stocks to a depth of 30 cm for different soil type- vegetation associations in Brazil. For example, the soil C stock in HAC soils under 14 different land cover categories, including Amazon forest and Brazilian Cerrado, range from 2 to 116 kg C m-2. Important Note: In most research that estimates carbon storage and sequestration rates on a landscape, soil pool measures only include soil organic carbon (SOC) in mineral soils (Post and Kwon 2000). However, if the ecosystem being modeled has a lot of organic soils (e.g. wetlands or paramo), it is critical to add this component to the mineral soil content. In landscapes where the conversion of wetlands into other land uses is common, carbon releases from organic soils should also be tracked closely (IPCC 2006).
14.7.6 2.4. Carbon stored in dead organic matter
If local or regional estimates of carbon stored in dead organic matter aren’t available, default values from the IPCC (2006) can be assigned. Table 2.2 (p. 2.27) gives default carbon stocks for leaf litter in forested LULC types. For non-forested types, litter is close to 0. Grace et al. (2006) estimate the average carbon stored in litter for major savanna ecosystems around the world (Table 1). It is not clear if their total “above-ground biomass” estimates include deadwood or not. Deadwood stocks are more difficult to estimate in general, and we have located no default data sources. Regional estimates:
14.7. Appendix: data sources 239 InVEST User Guide, Release 2.5.6
• United States: Smith et al. (2006) estimate carbon storage in litter (referred to as “Forest Floor” C in the document) and dead wood (the aggregate of C pools referred to as “Standing Dead Trees” and “Down Dead Wood” in the document) for all major forest types and forest management practices in each region of the U.S. as a function of stand age. • South America: Delaney et al. (1998) estimate carbon stored in standing and down dead wood in 6 tropical forests of Venezuela. According to the authors, deadwood is typically 1/10 the amount of biomass as above- ground vegetation.
14.7.7 3. Decay rates for harvested wood products
For more information on the decay of carbon in HWP and methods for estimating it, see Skog et al. (2004), Green et al. (2006), Miner (2006), Smith et al. (2006), chapter 12, “Harvested Wood Products,” of IPCC (2006), and Dias et al. (2007).
14.7.8 4. Harvest rates and dates harvest began
For an example of estimating carbon content in harvested wood products, we can use data from Makundi (2001). Assume that a softwood plantation in Tanzania has been producing timber for 50 years on a 5- hectare plot. Further, the rotation period for this type of plantation is 25 years (Makundi 2001). Assume an even age forestry operation. Therefore, every year, 2 hectares with 25-year old trees are clear-cut. The mean annual increment of the softwood’s aboveground biomass is 17.82 Mg ha-1 yr-1 (Makundi 2001). Thus 2 hectares x 25 years x 17.82 Mg ha-1 yr-1= 891 Mg of timber has been removedfrom the plantation annually for 50 years. If we assume the carbon content of the plantation’s trees are 0.48 (Makundi 2001) then 891 x 0.48 = 427.68 metric tons of C are in the aboveground biomass of forest stand removed each year from the plantation or 8.6 ha-1yr-1. Ascertaining dates in which harvesting began in each parcel may be difficult. If it is, you could assign an early date of initial harvest to all parcels, which essentially assumes that the carbon in the pool of harvested wood products has reached steady state (i.e., does not change year to year). Assume a date such that the time since first harvest is more than twice the half-life of carbon in the harvested wood products (e.g., if the half life of carbon in wood products is 20 years, choose a date of initial harvest that is 40 years before the current landscape map used.
14.7.9 5. Economic inputs: carbon price and discount rates
Recent estimates suggest that the social cost of carbon (SCC), or the marginal damage associated with the release of an additional Mg of C into the atmosphere, ranges from $32 per metric ton of C (Nordhaus 2007a) to $326 per metric ton of C (Stern 2007) in 2010 US dollars. The value of this damage can also be considered the monetary benefit of an avoided release. Tol (2009) provides a comprehensive survey of SCC estimates, reporting median values of $66 and $130 per metric ton in 2010 US dollars (values differ because of different assumptions regarding discounting of time). Other recent estimates can be found in Murphy et al. (2004), Stainforth et al. (2005), and Hope (2006). An alternative method for measuring the cost of an emission of a metric ton of C is to set the cost equal to the least cost alternative for sequestering that ton. The next best alternative currently is to capture and store the C emitted from utility plants. According to Socolow (2005) and Socolow and Pacala (2007), the cost of this technology per metric ton captured and stored is approximately $100. Finally, while we do not recommend this approach, market prices can be used to set the price of se- questered carbon. The Chicago Climate Exchange (CCX) and the European Climate Exchange (ECX) provide values ($24 and $153 per metric ton of C on May 14, 2008, respectively). The difference in these prices illustrates the problem with using markets to set values. The CCX and ECX are different in
14.7. Appendix: data sources 240 InVEST User Guide, Release 2.5.6
structure, scope, and the public policy that grounds each institution. This leads to different market funda- mentals, and different prices for reasons unrelated to the social value of carbon sequestration. We do not recommend the use of market prices because they usually only apply to “additional” carbon sequestration; sequestration above and beyond some baseline sequestration rate. Further, carbon credit values from car- bon markets such as the Chicago or European Climate Exchanges are largely a function of various carbon credit market rules and regulations and do not necessarily reflect the benefit to society of a sequestered ton of carbon. Therefore, correct use of market prices would require estimating a baseline rate for the land- scape of interest, mapping additional sequestration, and then determining which additional sequestration is eligible for credits according to market rules and regulations. If the user is specifically interested in such an analysis please contact the InVEST team on the message boards at http://invest.ecoinformatics.org We discount the value of future payments for carbon sequestration to reflect society’s preference for payments that occur earlier rather than later. The US Office of Management and Budget recommends a 7% per annum market discount rate for US-based projects (OMB 1992). Discount rates vary for other parts of the world. The Asian Development Bank uses a rate of 10% to 12% when evaluating projects (http://www2.adb.org/water/topics/dams/pdf/eco081.pdf). Canada and New Zealand recommend 10% for their projects (Abusah and de Bruyn 2007). Some economists believe that a market or consumption discount rate of 7% to 12% is too high when dealing with the climate change analysis. Because climate change has the potential to severely disrupt economies in the future, the preference of society to consume today at the expense of both climate stability in the future and future generations’ economic opportunities is seen as unethical by some (Cline 1992, Stern 2007). According to this argument, analyses of the effects of climate change on society and policies designed to reduce climate change should use low discount rates to encourage greater GHG emission mitigation and therefore compensate for the potentially severe damages incurred by future generations (e.g., r = 0.014 in Stern (2007)). Recent government policies in several countries have supported the use of a very low discount rate for certain long-term projects (Abusah and de Bruyn 2007). The carbon discount rate, which reflects the greater climatic impact of carbon sequestered immediately over carbon sequestered in the future, is discussed in Adams et al. (1999), Plantinga et al. (1999), Feng 2005, and Nelson et al. (2008).
14.8 Carbon 3.0 Beta
Currently we are working on the next generation platform of InVEST and deploying parts of it as prototype InVEST models. Carbon has a 3.0 prototype which can be found in the InVEST 3.0 Beta toolbox inside the InVEST 2.5.6 toolbox. Currently it is only supported in ArcGIS 10. New features to the 3.0 version include: • Parameters from previous runs are automatically loaded into the user interface. • Model is broken up into a biophysical and valuation section for users who only want to calculate stored/sequestered carbon or value a precomputed storage map. • Runtime of the model has been improved. • The resolution option has been removed. The carbon storage and valuation map is generated at the finest resolution of the inputs to the model. Please send feedback or bug reports to [email protected].
14.9 References
Abusah, Sam and Bruyn, Clinton de. 2007. Getting Auckland on Track: Public Transport and New Zealand’s Economic. Ministry of Economic Development Working Paper. Accessed at
14.8. Carbon 3.0 Beta 241 InVEST User Guide, Release 2.5.6
Adams, DM, RJ Alig, BA McCarl, et al. 1999. Minimum cost strategies for sequestering carbon in forests. Land Econ75: 360-374. Anderson, JR, EE Hardy, JT Roach, RE Witmer. A Land Use and Land Cover Classification System for Use with Re- mote Sensor Data. Washington, DC: United States Government Printing Office; 1976. Geological Survey Professional Paper 964. Antle, JM, and B. Diagana. 2003. Creating Incentives for the Adoption of Sustainable Agricultural Practices in Developing Countries: The Role of Soil Carbon Sequestration. American Journal of Agricultural Economics85:1178- 1184. Baer, SG, DJ Kitchen, JM Blair, and CW Rice. 2002. Changes in Ecosystem Structure and Function along a Chronose- quence of Restored Grasslands. Ecological Applications 12:1688-1701. Bernoux, M., MDS Carvalho, B. Volkoff, and CC Cerri. 2002. Brazil’s soil carbon stocks. Soil Science Society of America Journal66:888-896. Brown, SL, PE Schroeder and JS Kern. Spatial distribution of biomass in forests of the eastern USA.Forest Ecology and Management 123 (1999: 81-90. Brown, S. 2002. Measuring carbon in forests: current status and future challenges. Environmental Pollution116:363- 372. Brown, S. Estimating Biomass and Biomass Change of Tropical Forests: a Primer. FAO Forestry Department; 1997. Report for FAO Forestry Paper 134. Brown, S. and PE Schroeder. 1999. Spatial patterns of aboveground production and mortality of woody biomass for eastern US forests. Ecological Applications9:968-980. Cairns, MA, PK Haggerty, R. Alvarez, BHJ De Jong, and I. Olmsted. 2000. Tropical Mexico’s recent land-use change: A region’s contribution to the global carbon cycle. Ecological Applications 10:1426-1441. Cairns, MA, S. Brown, EH Helmer, and GA Baumgardner. 1997. Root biomass allocation in the world’s upland forests. Oecologia111:1-11. Canadell, JG and MR Raupach. 2008. Managing Forests for Climate Change Mitigation. Science320:1456-1457. Cline, WR. 1992. The economics of global warming. Instuitute for International Economics, Washington, D.C. Coomes, DA, RB Allen, NA Scott, C. Goulding, and P. Beets. 2002. Designing systems to monitor carbon stocks in forests and shrublands. Forest Ecology and Management164:89-108. Conte, MN and MJ Kotchen. 2010. Explaining the price of voluntary carbon offsets. Climate Change Economics 1 (2):93-111. Capoor, K., and P. Ambrosi. State and Trends of the Carbon Market 2008. Washington, D.C.: World Bank Institute, 2008 May. Delaney, M., S. Brown, AE Lugo, A. Torres-Lezama, and NB Quintero. 1998. The quantity and turnover of dead wood in permanent forest plots in six life zones of Venezuela. Biotropica30:2-11. Detwiler, RP. 1986. Land Use Change and the Global Carbon Cycle: The Role of Tropical Soils. Biogeochemistry2:67-93. Dias, AC, M. Louro, L. Arroja, and I. Capela. 2007. Carbon estimation in harvested wood products using a country- specific method: Portugal as a case study. Environmental Science & Policy 10 (3):250-259. Edinburgh Centre for Carbon Management. The Establishing Mechanisms for Payments for Carbon Environmental Services in the Eastern Arc Mountains, Tanzania; 2007 May 2007. Fargione, J., J. Hill, D. Tilman, S. Polasky, and P. Hawthorne. 2008. Land Clearing and the Biofuel Carbon Debt. Science319:1235-1238.
14.9. References 242 InVEST User Guide, Release 2.5.6
Feng, H. 2005. The dynamics of carbon sequestration and alternative carbon accounting, with an application to the upper Mississippi River Basin. Ecological Economics54:23-35. Gaston, G., S. Brown, M. Lorenzini, and KD Singh. 1998. State and change in carbon pools in the forests of tropical Africa. Global Change Biology4:97-114. Glenday, J. 2006. Carbon storage and emissions offset potential in an East African tropical rainforest. Forest Ecology and Management235:72-83. Grace, J., J. San Jose, P. Meir, HS Miranda, and RA Montes. 2006. Productivity and carbon fluxes of tropical savannas. Journal of Biogeography33:387-400. Green, C, V. Avitabile, EP Farrell, and KA Byrne. 2006. Reporting harvested wood products in national greenhouse gas inventories: Implications for Ireland. Biomass and Bioenergy 30(2): 105-114. Gibbs, HK, S Brown, JO Niles, and JA Foley. 2007. Monitoring and estimating tropical forest carbon stocks: making REDD a reality. Environmental Research Letters2:045023. Hamilton, K., M Sjardin, T Marcello, and G Xu. Forging a Frontier: State of the Voluntary Carbon Markets 2008. Washington, D.C.: Ecosystem Marketplace and New Carbon Finance; 2008. Hope, CW. 2006. The social cost of carbon: what does it actually depend on? Climate Policy 6: 565–572 Houghton, RA. 2005. Tropical deforestation as a source of greenhouse gas emissions. In: Tropical Deforestation and Climate Change, Moutinho and Schwartzman [eds.]. Instituto de Pesquisa Ambiental da Amazonia and Environmental Defense, Belem,Brazil. Houghton, RA, and JL Hackler. 2006. Emissions of carbon from land use change in sub-Saharan Africa. Journal of Geophysical Research111. The Intergovernmental Panel on Climate Change (IPCC). 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use. Prepared by the National Greenhouse Gas Inventories Programme, Eggleston, HS, L. Buendia, K. Miwa, T. Ngara, and K. Tanabe (eds). Institute for Global Environmental Strategies (IGES), Hayama, Japan.
14.9. References 243 InVEST User Guide, Release 2.5.6
Munishi, PKT and TH Shear. 2004. Carbon Storage in Afromontane Rain Forests of the Eastern Arc Mountains of Tanzania: their Net Contribution to Atmospheric Carbon. Journal of Tropical Forest Science16:78-93. Murphy, JMet al. 2004. Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature 430, 768–772. Murray, B., B. Sohngen, and M. Ross. 2007. Economic consequences of consideration of permanence, leakage and additionality for soil carbon sequestration projects. Climatic Change80:127-143. Nascimento, HEM, and WF Laurance. 2002. Total aboveground biomass in central Amazonian rainforests: a landscape-scale study. Forest Ecology and Management168:311-321. Nelson, E., G. Mendoza, J. Regetz, S. Polasky, H. Tallis, D. Cameron, K. Chan, G. Daily, J. Goldstein, P. Kareiva, E. Lonsdorf, R. Naidoo, TH Ricketts, and R. Shaw. 2008. Modeling Multiple Ecosystem Services and Tradeoffs at Landscape Scales. Frontiers in Ecology and the EnvironmentForthcoming. Nordhaus, W. 2007a. Critical Assumptions in the Stern Review on Climate Change. Science 317 (5835): 201–202. Nordhaus, W. 2007b. A Review of the Stern Review on the Economics of Global Warming. Journal of Economic Literature 45: 686-702. Pagiola, S. 2008. Payments for environmental services in Costa Rica. Ecological Economics 65 (4): 712-724. Plantinga, AJ, and RA Birdsey. 1994. Optimal Forest Stand Management When Benefits are Derived from Carbon. Natural Resource Modeling 8(4): 373-387. Polasky, S, E Nelson, D Pennington, and K Johnson. 2010. The Impact of Land-Use Change on Ecosystem Services, Biodiversity and Returns to Landowners: A Case Study in the State of Minnesota. Environmental and Resource Economics, in press. Post, WM, WR Emanuel, PJ Zinke, and AG Stangenberger. 1982. Soil carbon pools and world life zones. Nature298:156-159. Post, WM, KC Kwon. 2000. Soil carbon sequestration and land-use change: processes and potential. Global Change Biology6:317-327. Raich, JW, AE Russell, K. Kitayama, WJ Parton, and PM Vitousek. 2006. Temperature influences carbon accumula- tion in moist tropical forests. Ecology87:76-87. Ruesch A, and HK Gibbs. 2008. New IPCC tier-1 global biomass carbon map for the year 2000. Avail- able:http://cdiac.ornl.gov/epubs/ndp/global_carbon/carbon_documentation.html. Accessed 2008 Jul 7. Schuman, GE, HH Janzen, and JE Herrick. 2002. Soil carbon dynamics and potential carbon sequestration by range- lands. Environmental Pollution, 116:391-396. Sedjo, RA and B. Sohngen. Carbon Credits for Avoided Deforestation. Washington, DC: Resources for the Future; 2007 October 2007. Report for RFF DP 07-47. Silver, WL, R. Ostertag, and AE Lugo. 2000. The potential for carbon sequestration through reforestation of aban- doned tropical agricultural and pasture lands. Restoration Ecology8:394-407. Skog, KE, K. Pingoud, and JE Smith. 2004. Method Countries Can Use to Estimate Changes in Carbon Stored in Harvested Wood Products and the Uncertainty of Such Estimates. Environmental Management 33, Supplement 1: S65–S73. Smith, JE, LS Heath, KE Skog, RA Birdsey. Methods for Calculating Forest Ecosystem and Harvested Carbon with Standard Estimates for Forest Types of the United States. Newtown Square, PA: US Department of Agriculture, Forest Service, Northeastern Research Station; 2006. Report for NE-343. Socolow, RH. 2005. Can We Bury Global Warming? Scientific American 293: 49-55. Socolow, RH and SW Pacala. 2006. A Plan to Keep Carbon in Check. Scientific American 295: 50-57.
14.9. References 244 InVEST User Guide, Release 2.5.6
Sohngen, Brent, RH Beach, and Kenneth Andrasko. 2008. Avoided Deforestation as a Greenhouse Gas Mitigation Tool: Economic Issues. Journal of Environmental Quality 37: 1368-1375. Stainforth, DA et al., 2005. Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature 433, 403–406. Stern, N. 2007. The Economics of Climate Change: The Stern Review. Cambridge and New York: Cambridge University Press. Tiessen, H., C. Feller, EVSB Sampaio, and P. Garin. 1998. Carbon Sequestration and Turnover in Semiarid Savannas and Dry Forest. Climatic Change40:105-117. Tilman, D., J. Hill, and C. Lehman. 2006. Carbon-Negative Biofuels from Low-Input High-Diversity Grassland Biomass. Science314:1598-1600. Tol, RSJ. 2005. The marginal damage costs of carbon dioxide emissions: an assessment of the uncertainties. Energy Policy33:2064-2074. Tol, RSJ. 2009. The Economic Effects of Climate Change.Journal of Economic Perspectives23: 29–51. USOMB (US Office of Management and Budget). 1992. Guidelines and Discount Rates for Benefit-Cost Analy- sis of Federal Programs Circular No. A-94 (Revised). Transmittal Memo No. 64. Washington DC: US Office of Management and Budget. Vagen, TG, R Lal, and BR Singh. 2005. Soil carbon sequestration in sub-Saharan Africa: A review. Land Degradation & Development16:53-71. Weitzman, ML. 2007. A review of the Stern Review on the Economics of Climate Change. Journal of Economic Literature45:703-724. Zhang, Q, and CO Justice. 2001. Carbon Emissions and Sequestration Potential of Central African Ecosystems. AMBIO30:351-355.
14.9. References 245 CHAPTER FIFTEEN
RESERVOIR HYDROPOWER PRODUCTION
15.1 Summary
Hydropower accounts for twenty percent of worldwide energy production, most of which is generated by reservoir systems. InVEST estimates the annual average quantity and value of hy- dropower produced by reservoirs, and identifies how much water yield or value each part of the landscape contributes annually to hydropower production. The model has three components: wa- ter yield, water consumption, and hydropower valuation. The first two components use data on average annual precipitation, annual reference evapotranspiration and a correction factor for vegetation type, root restricting layer depth, plant available water content, land use and land cover, root depth, elevation, saturated hydraulic conductivity, and consumptive water use. The valua- tion model uses data on hydropower market value and production costs, the remaining lifetime of the reservoir, and a discount rate. The biophysical models do not consider surface – ground water interactions or the temporal dimension of water supply. The val- uation model assumes that energy pricing is static over time.
15.2 Introduction
The provision of fresh water is an environmental service that con- tributes to the welfare of society in many ways, including through the production of hydropower, the most widely used form of renewable energy in the world. Most hydropower pro- duction comes from watershed-fed reservoir systems that generally deliver energy consistently and predictably. The systems are designed to account for annual variability in water volume, given the likely levels for a given watershed, but are vulnerable to extreme variation caused by land use and land cover (LULC) changes. LULC changes can alter hydrologic cycles, affecting patterns of evapotranspiration, infiltration and water retention, and changing the timing and volume of water that is available for hydropower production (World Commission on Dams 2000; Ennaanay 2006). Changes in the landscape that affect annual average water yield upstream of hydropower facilities can increase or decrease hydropower production capacity. Maps of where water yield used for hydropower is produced can help avoid unintended impacts on hydropower production or help direct land use decisions that wish to maintain power production, while balancing other uses such as conservation or agriculture. Such maps can also be used to inform
246 InVEST User Guide, Release 2.5.6 investments in restoration or management that downstream stakeholders, such as hydropower companies, make in hopes of improving or maintaining water yield for this important environmental service. In large watersheds with multiple reservoirs for hydropower production, areas upstream of power plants that sell to a higher value market will have a higher value for this service. Maps of how much value each parcel contributes to hydropower production can help managers avoid developments in the highest hydropower value areas, understand how much value will be lost or gained as a consequence of different management options, or identify which hydropower producers have the largest stake in maintaining water yield across a landscape.
15.2.1 Reservoir Hydropower Production 3.0 Beta
We are working on the next generation platform of InVEST and deploying parts of it as prototype InVEST models. Reservoid Hydropower Production has a 3.0 prototype which can be found in the Windows Start menu after the InVEST installation is complete. New features to the 3.0 version include: • Large performance improvements to the runtime of the model. • Outputs are simplified into shapefile polygons rather than rasterized polygons. Generally the raster outputs of the ArcGIS versions of the models have a field in a shapefile that corresponds to that output. • The ArcGIS model is run in 3 separate steps. The standalone model has a streamlined interface to run in a single step.
15.3 The Model
The InVEST Reservoir Hydropower model estimates the relative contributions of water from different parts of a landscape, offering insight into how changes in land use patterns affect annual surface water yield and hydropower production. Modeling the connections between landscape changes and hydrologic processes is not simple. Sophisticated models of these connections and associated processes (such as the WEAP model) are resource and data intensive and require substantial expertise. To accommodate more contexts, for which data is readily available, InVEST maps and models the annual average water yield from a landscape used for hydropower production, rather than directly addressing the affect of LULC changes on hydropower failure as this process is closely linked to variation in water inflow on a daily to monthly timescale. Instead, InVEST calculates the relative contribution of each land parcel to annual average hydropower production and the value of this contribution in terms of energy production. The net present value of hydropower production over the life of the reservoir also can be calculated by summing discounted annual revenues.
15.3.1 How it works
The model runs on a gridded map. It estimates the quantity and value of water used for hydropower production from each sub-basin in the area of interest. It has three components, which run sequentially. First, it determines the amount of water running off each pixel as the precipitation less the fraction of the water that undergoes evapotranspiration. The model does not differentiate between surface, subsurface and baseflow, but assumes that all water yield from a pixel reaches the point of interest via one of these pathways. This model then sums and averages water yield to the sub- basin level. The pixel-scale calculations allow us to represent the heterogeneity of key driving factors in water yield such as soil type, precipitation, vegetation type, etc. However, the theory we are using as the foundation of this set of models was developed at the sub-watershed to watershed scale. We are only confident in the interpretation of these models at the sub-watershed scale, so all outputs are summed and/or averaged to the sub-basin scale. We do continue to provide pixel-scale representations of some outputs for calibration and model-checking purposes only. These pixel- scale maps are not to be interpreted for understanding of hydrological processes or to inform decision making of any kind.
15.3. The Model 247 InVEST User Guide, Release 2.5.6
Second, beyond annual average runoff, it calculates the proportion of surface water that is used for hydropower pro- duction by subtracting the surface water that is consumed for other uses. Third, it estimates the energy produced by the water reaching the hydropower reservoir and the value of this energy over the reservoir’s lifetime.
Water Yield Model
The water yield model is based on the Budyko curve and annual average precipitation. First, we determine annual water yield Y (x) for each pixel on the landscape x as follows:
AET (x) Y (x) = 1 − · P (x) P (x)
where, `x is the land cover type for pixel x, AET (x) is the annual actual evapotranspiration for pixel x and P (x) is the annual precipitation on pixel x.
Figure 1. Conceptual diagram of the water balance model used in the hydropower production model. The water cycle is simplified, including only the parameters shown in color, and ignoring the parameters shown in gray. Yield, as calculated by this step of the model, is then adjusted for other consumptive uses and applied to hydropower energy and value estimates. AET (x) The evapotranspiration partition of the water balance, P (x) , is an approximation of the Budyko curve developed by Zhang et al. (2001):
AET (x) 1 + ω(x)R(x) = 1 P (x) 1 + ω(x)R(x) + R(x)
15.3. The Model 248 InVEST User Guide, Release 2.5.6
where, R(x) is the dimensionless Budyko Dryness index on pixel x, defined as the ratio of potential evapotranspiration to precipitation (Budyko 1974) and ω(x) is a modified dimensionless ratio of plant accessible water storage to expected precipitation during the year. As defined by Zhang et al. (2001), ω(x) is a non-physical parameter to characterize the natural climatic-soil properties.
AW C(x) ω(x) = Z P (x)
where AW C(x) is the volumetric (mm) plant available water content. The soil texture and effective rooting depth define AW C(x), which establishes the amount of water that can be held and released in the soil for use by a plant, estimated as the product of the difference between field capacity and wilting point and the minimum of root restricting layer depth and vegetation rooting depth. Root restricting layer depth is the soil depth at which root penetration is strongly inhibited because of physical or chemical characteristics. Vegetation rooting depth is often given as the depth at which 95% of a vegetation type’s root biomass occurs. Z is a seasonality factor that presents the seasonal rainfall distribution and rainfall depths. In areas of winter rains, we expect to have Z on the order of 10, in humid areas with rain events distributed throughout the year or regions with summer rains the Z is on the order of 1. While we calculate ω(x), in some cases specific biome values already exist based on water availability and soil-water storage (Milly 1994, Potter et al. 2005, Donohue et al. 2007). Finally, we define the Budyko dryness index, where R(x) values that are greater than one denote pixels that are potentially arid (Budyko 1974), as follows:
K (` ) · ET (x) R(x) = c x 0 P (x)
where, ET0(x) is the reference evapotranspiration from pixel x and Kc(`x) is the plant (vegetation) evapotranspiration coefficient associated with the LULC `x on pixel x. ET0(x) reflects local climatic conditions, based on the evapo- transpiration of a reference vegetation such as grass of alfalfa grown at that location. Kc(`x) is largely determined by the vegetative characteristics of the land use/land cover found on that pixel (Allen et al. 1998). Kc adjusts the ET0 values to the crop or vegetation type in each pixel of the land use/land cover map. Kc adjusts the ET0 values to the crop or vegetation type in each pixel of the land use/land cover map, and is then used to estimate actual ET (AET) for the watershed, one of the model outputs. The water yield model script generates and outputs the total and average water yield at the sub-basin level.
Water Scarcity Model
The Water Scarcity Model calculates the water scarcity value based on water yield and water consumptive use in the watershed(s) of interest. The user inputs how much water is consumed by each land use land cover type in a table format. For example, in an urban area, consumptive use can be calculated as the product of population density and per capita consumptive use. These land use-based values only relate to the consumptive portion of demand; some water use is non-consumptive such water used for cooling or other industrial processes that return water to the stream after use. For simplicity, each pixel in the watershed is either a “contributing” pixel, which contributes to hydropower production, or a “use” pixel, which uses water for other consumptive uses. This assumption implies that land use associated with consumptive uses will not contribute any yield for downstream use. The amount of water that actually reaches the reservoir for dam d (realized supply) is defined as the difference between total water yield from the watershed and total consumptive use in the watershed.
Vin = Y − ud
where ud is the total volume of water consumed in the watershed upstream of dam d and Y is the total water yield from the watershed upstream of dam d.
15.3. The Model 249 InVEST User Guide, Release 2.5.6
If the user has observed data available on actual annual inflow rates to the reservoir for dam d, they can be compared to Vin. Divide the observed value by the estimated value to derive a calibration constant. This can then be entered in to the hydropower calibration table and used to make power and value estimates actual rather than relative.
Hydropower Production and Valuation Model
The reservoir hydropower model estimates both the amount of energy produced given the estimated realized supply of water for hydropower production and the value of that energy. A present value dollar (or other currency) estimate is given for the entire remaining lifetime of the reservoir. Net present value can be calculated if hydropower produc- tion cost data are available. The energy produced and the revenue is then redistributed over the landscape based on the proportional contribution of each sub-watershed to energy production. Final output maps show how much en- ergy production and hydropower value can be attributed to each sub-watershed’s water yield over the lifetime of the reservoir. At dam d, power is calculated using the following equation:
pd = ρ · qd · g · hd
3 3 where pd is power in watts, ρ is the water density (1000 Kg/m ), qd is the flow rate (m /s), g is the gravity constant 2 (9.81 m/s ), and hd is the water height behind the dam at the turbine (m). In this model, we assume that the total annual inflow water volume is released equally and continuously over the course of each year. The power production equation is connected to the water yield model by converting the annual inflow volume adjusted for consumption (Vin) to a per second rate. Since electric energy is normally measured in kilowatt-hours, the power pd is multiplied by the number of hours in a year. All hydropower reservoirs are built to produce a maximum amount of electricity. This is called the energy production rating, and represents how much energy could be produced if the turbines are 100% efficient and all water that enters the reservoir is used for power production. In the real world, turbines have inefficiencies and water in the reservoir may be extracted for other uses like irrigation, retained in the reservoir for other uses like recreation, or released from the reservoir for non-power production uses like maintaining environmental flows downstream. To account for these inefficiencies and the flow rate and power unit adjustments, annual average energy production εd at dam d is calculated as follows:
εd = 0.00272 · β · γd · hd · Vin
where εd is hydropower energy production (KWH), β is the turbine efficiency coefficient (%), γd is the percent of inflow water volume to the reservoir at dam d that will be used to generate energy.
To convert εd, the annual energy generated by dam d, into a net present value (NPV) of energy produced (point of use value) we use the following,
T −1 X 1 NPVH = (p ε − TC ) × d e d d (1 + r)t t=0
where TCd is the total annual operating costs for dam d, pe is the market value of electricity (per unit of energy consumed) provided by hydropower plant at dam d, Td indicates the number of years present landscape conditions are expected to persist or the expected remaining lifetime of the station at dam d (set T to the smallest value if the two time values differ), and r is the market discount rate. The form of the equation above assumes that TCd, pe, and εd, are constant over time. Energy production over the lifetime of dam d is attributed to each sub-watershed as follows:
εx = (Tdεd) × (cx/ctot)
15.3. The Model 250 InVEST User Guide, Release 2.5.6 where the first term in parentheses represents the electricity production over the lifetime of dam d. The second term represents the proportion of water volume used for hydropower production that comes from sub-watershed x relative to the total water volume for the whole watershed. The value of each sub-watershed for hydropower production over the lifetime of dam d is calculated similarly:
NPVHx = NPVHd × (cx/ctot)
Limitations and simplifications
The model has a number of limitations. First, it is not intended for devising detailed water plans, but rather for evaluating how and where changes in a watershed may affect hydropower production for reservoir systems. It is based on annual averages, which neglect extremes and do not consider the temporal dimensions of water supply and hydropower production. Second, the model assumes that all water produced in a watershed in excess of evapotranspiration arrives at the watershed outlet, without considering water capture by means other than primary human consumptive uses. Surface water – ground water interactions are entirely neglected, which may be a cause for error especially in areas of karst geology. The relative contribution of yield from various parts of the watershed should still be valid. Third, the model does not consider sub-annual patterns of water delivery timing. Water yield is a provisioning function, but hydropower benefits are also affected by flow regulation. The timing of peak flows and delivery of minimum operational flows throughout the year determines the rate of hydropower production and annual revenue. Changes in landscape scenarios are more likely to affect the timing of flows than the annual water yield, and are more of a concern when considering drivers such as climate change. Modeling the temporal patterns of overland flow requires detailed data that are not appropriate for our approach. Still, this model provides a useful initial assessment of how landscape scenarios may affect the annual delivery of water to hydropower production. Fourth, the model describes consumptive demand by LULC type. In reality, water demand may differ greatly between parcels of the same LULC class. Much of the water demand may also come from large point source intakes, which are not represented by LULC class. The model simplifies water demand by distributing it over the landscape. For example, the water demand may be large for an urban area, and the model represents this demand by distributing it over the urban LULC class. The actual water supply intake, however, is likely much further upstream in a rural location. Spatial disparity in actual and modeled demand points may cause an incorrect representation in the scarcity output grid. The distribution of consumption is also simplified in the reallocation of energy production and hydropower value since it is assumed that water consumed along flow paths is drawn equally from every pixel upstream. As a result, water scarcity, energy production patterns, and hydropower values may be incorrectly estimated.
Fifth, a single variable (γd) is used to represent multiple aspects of water resource allocation, which may misrepresent the complex distribution of water among uses and over time. Finally, the model assumes that hydropower production and pricing remain constant over time. It does not account for seasonal variation in energy production or fluctuations in energy pricing, which may affect the value of hydropower. Even if sub-annual production or energy prices change, however, the relative value between parcels of land in the same drainage area should be accurate.
15.4 Data needs
Here we outline the specific data used by the model. See the appendix for detailed information on data sources and pre-processing. For all raster inputs, the projection used should be defined, and the projection’s linear units should be in meters. The following link is a concise presentation of the naming conventions and properties of the different data types input into this model http://won.sdsu.edu/protected130/evapotranspiration_studies_1110230800.html.
15.4. Data needs 251 InVEST User Guide, Release 2.5.6
1. Root restricting layer depth (required). A GIS raster dataset with an average root restricting layer depth value for each cell. Root restricting layer depth is the soil depth at which root penetration is strongly inhibited because of physical or chemical characteristics. The root restricting layer depth values should be in millimeters. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with an average root restricting layer depth in millimeters for each cell. Sample data set: \InVEST\Base_Data\Freshwater\soil_depth 2. Precipitation (required). A GIS raster dataset with a non-zero value for average annual precipitation for each cell. The precipitation values should be in millimeters. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with precipitation values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\precip 3. Plant Available Water Content (required). A GIS raster dataset with a plant available water content value for each cell. Plant Available Water Content fraction (PAWC) is the fraction of water that can be stored in the soil profile that is available for plants’ use. PAWC is a fraction from 0 to 1. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with available water content values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\pawc 4. Average Annual Reference Evapotranspiration (required). A GIS raster dataset, with an annual average evapotranspiration value for each cell. Reference evapotranspiration is the potential loss of water from soil by both evaporation from the soil and transpiration by healthy alfalfa (or grass) if sufficient water is available. The reference evapotranspiration values should be in millimeters. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with reference evapotranspiration values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\eto 5. Land use/land cover (required). A GIS raster dataset, with an LULC code for each cell. The LULC code should be an integer. Name: File can be named anything, but no spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with an integer LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.). These codes must match LULC codes in the Biophysical table. Sample data set: \InVEST\Base_Data\Freshwater\landuse_90 6. Watersheds (required). A shapefile, with one polygon per watershed. This is a layer of watersheds such that each watershed contributes to a point of interest where hydropower production will be analyzed. See the Working with the DEM section for information about generating watersheds. Name: File can be named anything, but no spaces in the name Format: Shapefile (.shp) Rows: Each row is one watershed Columns: An integer field named ws_id is required, with a unique integer value for each watershed
15.4. Data needs 252 InVEST User Guide, Release 2.5.6
Sample data set: \InVEST\Base_Data\Freshwater\watersheds.shp 7. Sub-watersheds (required). A shapefile, with one polygon per sub-watershed within the main watersheds specified in the Watersheds shapefile. See the Working with the DEM section for information about generating sub-watersheds. Format: Shapefile (.shp) Rows: Each row is one sub-watershed Columns: An integer field named subws_id is required, with a unique integer value for each sub-watershed Sample data set: \InVEST\ Base_Data\Freshwater\subwatersheds.shp 8. Biophysical Table (required). A table of land use/land cover (LULC) classes, containing data on biophysical coefficients used in this tool. NOTE: these data are attributes of each LULC class rather than attributes of individual cells in the raster map. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Biophysical_Models Name: Table names should only have letters, numbers and underscores, no spaces Format: *.dbf or *.mdb for ArcGIS models, the standalone model requires a .csv file Rows: Each row is an LULC class. Columns: Each column contains a different attribute of each land use/land cover class, and must be named as follows: 1. lucode (Land use code): Unique integer for each LULC class (e.g., 1 for forest, 3 for grassland, etc.), must match the LULC raster above. 2. LULC_desc: Descriptive name of land use/land cover class (optional) 3. LULC_cat: Contains one of the following four categories: “water”, “wetlands”, “built”, and “veg”. Both standing and flowing water bodies should be assigned to the “water” category. Urban and paved areas should be assigned to built. All areas that are not water, wetland or built should be assigned to “veg”. This is used to determine which function is used to calculate AET. 4. root_depth: The maximum root depth for vegetated land use classes, given in integer millimeters. This is often given as the depth at which 95% of a vegetation type’s root biomass occurs. We apply different equations for a few special cases where the generic Budyko curve approach is not appropriate. In these cases, the rooting depth should be set to NA. 5. :math:‘K_c‘: The plant evapotranspiration coefficient for each LULC class, used to obtain potential evapotranspiration by using plant physiological characteristics to modify the reference evapotran- spiration, which is based on alfalfa. Coefficients should be multiplied by 1000, so that the final Kc values given in the table are integers ranging between 1 and 1500. (Some crops evapotranspire more than alfalfa in some very wet tropical regions and where water is always available). 9. seasonality factor (Z) (required). Floating point value on the order of 1 to 10 corresponding to the seasonal distribution of precipitation (see Appendix A for more information). 10. Demand Table (required). A table of LULC classes, showing consumptive water use for each landuse / land- cover type. Consumptive water use is that part of water used that is incorporated into products or crops, con- sumed by humans or livestock, or otherwise removed from the watershed water balance. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Water_Demand Name: Table names should only have letters, numbers and underscores, no spaces Format: *.dbf or *.mdb Rows: Each row is a landuse / landcover class
15.4. Data needs 253 InVEST User Guide, Release 2.5.6
Columns: Contain water demand values per LULC class and must be named as follows: 1. lucode: Integer value of land use/land cover class (e.g., 1 for forest, 3 for grassland, etc.), must match LULC raster, described above. 2. demand: The estimated average consumptive water use for each landuse / landcover type. Water use should be given in cubic meters per year for a pixel in the land use/land cover map. Note that accounting for pixel area is important since larger areas will consume more water for the same land cover type. 11. Hydropower valuation table. A table of hydropower stations with associated model values. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Hydropower_Valuation Name: Table names should only have letters, numbers and underscores, no spaces Format: *.dbf or *.mdb Rows: Each row is a hydropower station Columns: Each column contains an attribute of each hydropower station, and must be named as follows: 1. ws_id: Unique integer value for each watershed, which must correspond to values in the Watersheds layer. 2. station_desc: Name of hydropower station (optional) 3. efficiency: The turbine efficiency. A number to be obtained from the hydropower plant manager (floating point values generally 0.7 to 0.9) 4. fraction: The fraction of inflow water volume that is used to generate energy, to be obtained from the hydropower plant manager. Managers can release water without generating electricity to satisfy irrigation, drinking water or environmental demands. Floating point value. 5. height: The head, measured as the average annual effective height of water behind each dam at the turbine intake in meters. Floating point value. 6. kw_price: The price of one kilowatt-hour of power produced by the station, in dollars or other currency. Floating point value. 7. cost: Annual cost of running the hydropower station (maintenance and operations costs). Floating point value. 8. time_span: An integer value of either the expected lifespan of the hydropower station or the period of time of the land use scenario of interest. Used in net present value calculations. 9. discount: The discount rate over the time span, used in net present value calculations. Floating point value. 12. Hydropower calibration table. A table of hydropower stations with associated calibration values. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Hydropower_Calibration Name: Table names should only have letters, numbers and underscores, no spaces Format: *.dbf or *.mdb Rows: Each row is a hydropower station Columns: Each column contains an attribute of each hydropower station, and must be named as follows: 1. ws_id: Unique integer value for each watershed, which must correspond to values in the Watersheds layer. 2. calib: Annual water yield calibration constant. Multiplying this value by the total water supply for a watershed should give the actual total annual water supply observed/measured at the point of interest,
15.4. Data needs 254 InVEST User Guide, Release 2.5.6
corresponding to the cyield column of the Scarcity tool’s water_scarcity.dbf output. Floating point value.
15.5 Running the Model
The Hydropower model maps the water yield, water consumption, energy produced by water yield and corresponding energy value over the landscape. This model is structured as a toolkit which has three tools. The first tool, Water Yield, calculates the surface water yield and actual evapotranspiration across the landscape. This output feeds into the next portion of the model, the Water Scarcity tool, which calculates water consumption, supply and realized supply, which is yield minus consumption. The third tool, Valuation, calculates energy production and the value of that energy, as it can be attributed to sub-basins on the watershed of interest. By running the tool, three folders will automatically be created in your workspace (you will have the opportunity to define this file path), “Intermediate”, where temporary files are written, and which is deleted after each tool run; “Service”, where results that show environmental services are saved; and “Output”, where non-service biophysical results are saved. Before running the Hydropower Model, make sure that the InVEST toolbox has been added to your ArcMap document, as described in the Getting Started chapter of this manual. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. • Identify workspace If you are using your own data, you need to first create a workspace, or folder for the analysis data, on your computer hard-drive. The entire pathname to the workspace should not have any spaces. All your output files will be saved here. For simplicity, you may wish to call the folder for your workspace ‘Hydropower’ and create a folder in your workspace called “Input” and place all your input files here. It’s not necessary to place input files in the workspace, but advisable so you can easily see the data you use to run your model. Or, if this is your first time using the tool and you wish to use sample data, you can use the data provided in InVEST- Setup.exe. If you installed the InVEST files on your C drive (as described in the Getting Started chapter), you should see a folder called /InVEST/Hydropower. This folder will be your workspace. The input files are in a folder called /InVEST/Base_Data/Freshwater. • Open an ArcMap document to run your model. • Find the InVEST toolbox in ArcToolbox. ArcToolbox is normally open in ArcMap, but if it is not, click on the ArcToolbox symbol. See the Getting Started chapter if you don’t see the InVEST toolbox and need instructions on how to add it. • You can run this analysis without adding data to your map view, but usually it is recommended to view your data first and get to know them. Add the data for this analysis to your map using the ADD DATA button and look at each file to make sure it is formatted correctly. Save your ArcMap file as needed. • Click once on the plus sign on the left side of the InVEST toolbox to see the list of tools expand. Next, click on the plus sign next to the Hydropower toolset. Within the toolset are three tools, Water Yield, Water Scarcity and Valuation. You will need to run Water Yield first, Water Scarcity second and Valuation last, as each tool generates outputs that feed into the next. • Double click on Water Yield. An interface will pop up like the one below. The tool shows default file names, but you can use the file buttons to browse instead to your own data. When you place your cursor in each space, you can read a description of the data requirements in the right side of the interface. Click Show Help if the description is not displayed. In addition, refer to the Data Needs section above for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires you to enter some data. • After you’ve entered all values as required, click on OK. The script will run, and its progress will be indicated by a “Progress dialogue”.
15.5. Running the Model 255 InVEST User Guide, Release 2.5.6
15.5. Running the Model 256 InVEST User Guide, Release 2.5.6
• Load the output grids into ArcMap using the ADD DATA button from either “Output” or “Service” folders. • You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting PROPERTIES, and then SYMBOLOGY. There are many options here to change the way the file appears in the map. You may change the coloring scheme for better visualization. • You can also view the attribute data of many output files by right clicking on a layer and selecting OPEN ATTRIBUTE TABLE. • Now, run the tool Water Scarcity. Several outputs from the Water Yield model, wyield_vol, wyield_mn (in the Service folder), water_yield_watershed.dbf and water_yield_subwatershed.dbf (in the Output folder) serve as inputs to this model (see results interpretation section). The interface is below:
• When the script completes running, its results will be saved in the Output folder. A description of these results is in the next section. Load them into your ArcMap project, look at them, and check out the attribute table. • Finally, run the tool Valuation. These outputs from the Water Scarcity tool are required: cyield_vol, con- sum_vol, water_scarcity_watershed.dbf and water_scarcity_subwatershed.dbf. The interface is below: • When the script completes running, its results will be saved in the Service folder. A description of these results is in the next section. Load them into your ArcMap project, look at them, and check out the attribute table. This model is open source, so you can edit the scripts to modify, update, and/or change equations by right clicking on
15.5. Running the Model 257 InVEST User Guide, Release 2.5.6
the script’s name and selecting “Edit...” The script will open in a text editor. After making changes, click File/Save to save your new script.
15.6 Interpreting Results
The following is a short description of each of the outputs from the Hydropower Production model. Final results are found in the Output and Service folders within the Workspace specified for this model. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix. • Output\fractp_mn (fraction): Mean actual evapotranspiration fraction of precipitation per sub-watershed (Ac- tual Evapotranspiration / Precipitation). It is the mean fraction of precipitation that actually evapotranspires at the sub-basin level. (this output is removed in the standalone model in favor of a shapefile output) • Output\aet_mn (mm): Mean actual evapotranspiration per sub-watershed. (this output is removed in the stan- dalone model in favor of a shapefile output) • Service\wyield_vol (m3): Total water yield per sub-watershed. The approximate absolute annual water yield across the landscape, calculated as the difference between precipitation and actual evapotranspiration on each land parcel. (this output is removed in the standalone model in favor of a shapefile output) • Service\wyield_mn (mm): Mean water yield per sub-watershed. (this output is removed in the standalone model in favor of a shapefile output) • Service\wyield_ha (m3/ha): Water yield volume per hectare per sub-watershed. (this output is removed in the standalone model in favor of a shapefile output)
15.6. Interpreting Results 258 InVEST User Guide, Release 2.5.6
• Output\water_yield_watershed.dbf: Table containing biophysical output values per watershed. (in the 3.0 version of this model, the output is a CSV file) With fields as follows: – precip_mn (mm): Mean precipitation per pixel on the watershed. – PET_mn (mm): Mean potential evapotranspiration per pixel on the watershed. – AET_mn (mm): Mean actual evapotranspiration per pixel on the watershed. – wyield_mn (mm): Mean water yield per pixel on the watershed. – wyield_vol (m3): Volume of water yield in the watershed. • Output\water_yield_subwatershed.dbf: Table containing biophysical output values per sub-watershed if a sub-watershed shapefile is provided. (in the 3.0 version of this model, the output is a CSV file) With fields as follows: – precip_mn (mm): Mean precipitation per pixel on the sub-watershed. – PET_mn (mm): Mean potential evapotranspiration per pixel on the sub-watershed. – AET_mn (mm): Mean actual evapotranspiration per pixel on the sub-watershed. – wyield_mn (mm): Mean water yield per pixel on the sub-watershed. – wyield_vol (m3): Volume of water yield in the sub-watershed. • Output\consum_vol (m3): Total water consumption for each sub-watershed. (this output is removed in the standalone model in favor of a shapefile output) • Output\consum_mn (m3/ha): Mean water consumptive volume per hectare per sub-watershed. (this output is removed in the standalone model in favor of a shapefile output) • Output\rsup_vol (m3): Total realized water supply (water yield – consumption) volume for each sub-watershed. (this output is removed in the standalone model in favor of a shapefile output) • Output\rsup_mn (m3/ha): Mean realized water supply (water yield – consumption) volume per hectare per sub-watershed. (this output is removed in the standalone model in favor of a shapefile output) • Output\cyield_vol (m3): Calibrated water yield volume per sub-watershed (water yield * calibration constant). (this output is removed in the standalone model in favor of a shapefile output) • Output\water_scarcity_watershed.dbf: Table containing values from water_yield_watershed.dbf, plus addi- tional fields as follows (in the 3.0 version of this model, the output is a CSV file): – cyield_vl (m3): Total calibrated water yield (water yield * calibration constant) per watershed. – consump_vl (m3): Total water consumption per watershed. – consump_mn (m3): Mean water consumption per watershed. – rsupply_vl (m3): Total realized water supply (water yield – consumption) per watershed. – rsupply_mn (m3): Mean realized water supply (water yield – consumption) per watershed. • Output\water_scarcity_subwatershed.dbf: Table containing values from water_yield_subwatershed.dbf, plus additional fields as follows (in the 3.0 version of this model, the output is a CSV file): – cyield_vl (m3): Total calibrated water yield (water yield * calibration constant) per sub-watershed. – consump_vl (m3): Total water consumption per sub-watershed. – consump_mn (m3): Mean water consumption per sub-watershed. – rsupply_vl (m3): Total realized water supply (water yield – consumption) per sub-watershed. – rsupply_mn (m3): Mean realized water supply (water yield – consumption) per sub-watershed.
15.6. Interpreting Results 259 InVEST User Guide, Release 2.5.6
• Service\hp_energy (kw/timespan): THIS IS THE SUB-WATERSHED MAP OF THIS ENVIRONMENTAL SERVICE IN ENERGY PRODUCTION TERMS. This grid shows the amount of energy produced by the hy- dropower station over the specified timespan that can be attributed to each sub-watershed based on its water yield contribution. (this output is removed in the standalone model in favor of a shapefile output) • Service\hp_val (currency/timespan): THIS IS THE SUB-WATERSHED MAP OF THIS ENVIRONMENTAL SERVICE IN ECONOMIC TERMS. This grid shows the value of the landscape per sub-watershed according to its ability to yield water for hydropower production over the specified timespan. (this output is removed in the standalone model in favor of a shapefile output) • Service\hydropower_value_watershed.dbf: Table containing values from water_scarcity_watershed.dbf, plus additional fields as follows (in the 3.0 version of this model, the output is a CSV file): – hp_energy (kw/timespan): Amount of energy produced over the specified timespan per watershed. – hp_value (currency/timespan): Value of the watershed landscape for producing energy over the specified timespan. • Service\hydropower_value_subwatershed.dbf: Table containing values from wa- ter_scarcity_subwatershed.dbf, plus additional fields as follows (in the 3.0 version of this model, the output is a CSV file): – hp_energy (kw/timespan): Amount of energy produced over the specified timespan per sub-watershed. – hp_value (currency/timespan): Value of the sub-watershed landscape for producing energy over the speci- fied timespan. • (3.0 only) Output\wyield_sub_sheds.shp, Output\wyield_sheds.shp, Output\scarcity_sub_sheds.shp, Out- put\scarcity_sheds.shp, Output\valuation_sub_sheds.shp, Output\valuation_sheds.shp: These files are shapefile equivalents of the pixel outputs from the ArcGIS models above, but rather than rasterizing the re- sults, users can symbolize the polygons directly thus simplifying data management and storage space. The application of these results depends entirely on the objective of the modeling effort. Users may be interested in all of these results or a select one or two. If costing information is not available or of interest, the user may choose to simply run the water yield model and compare biophysical results. The first several model results provide insight into how water is distributed through the landscape. aet_mn describes the actual evapotranspiration depth of the hydrologic cycle, showing how much water (precipitation) is lost annually to evapotranspiration across the watershed. The wyield_vol raster shows the annual average water volume that is ‘yielded’ from each sub-watershed of the wa- tershed of interest. This raster can be used to determine which sub-watersheds are most important to total annual water yield – although at this step the user still will not know how much of that water is benefiting downstream users of any type. The consumptive use (consum_vol) grid then shows how much water is used for consumptive activities (such as drinking, bottling, etc.) each year across the landscape. The realized supply (rsupply_vol) grid calculates the difference between cumulative water yield and cumulative consumptive use. This grid demonstrates where the water supply for hydropower production is abundant and where it is most scarce. The user needs to remember that the consumptive use grid may not truly represent where water is taken, only where it is demanded. This may cause some misrepresentation of the scarcity in certain locations, but this grid offers a general sense of the water balance and whether there is a lack of or abundance of water in the area of interest. The hp_energy and hp_val grids are the most relevant model outputs for prioritizing the landscape for investments that wish to maintain water yield for hydropower production. The hp_val grid contains the most information for this purpose as it represents the revenue attributable to each sub-watershed over the expected lifetime of the hydropower station, or the number of years that the user has chosen to model. This grid accounts for the fact that different hydropower stations within a large river basin may have different customers who pay different rates for energy pro- duction. If this is the case, this grid will show which sub-watersheds contribute the highest value water for energy production. If energy values do not vary much across the landscape, the hp_energy outputs can be just as useful in planning and prioritization. Comparing any of these grids between landuse scenarios allows the user to understand how the role of the landscape may change under different management plans.
15.6. Interpreting Results 260 InVEST User Guide, Release 2.5.6
The hydropower output summary tables present the model results in terms of hydropower operation. The cyield_vl field provides the total volume of water that arrives at each hydropower plant every year, considering water yield and consumption. The consump_vl field provides the total volume of water that is consumed in each watershed upstream of the station. Total energy produced at each hydropower station is given in the hp_energy field, and the corresponding value of that energy is given in the hp_value field. This table provides a quick comparison between land use scenarios in a way that complements the spatial representation across the landscape. Ideally the output grids and summary table will be used together for comparison of land use and management scenarios.
15.7 Appendix A: Data Sources
This is a rough compilation of data sources and suggestions about finding, compiling, and formatting data. This section should be used for ideas and suggestions only. We will continue to update this section as we learn about new data sources and methods. 1. Average annual precipitation Average Annual Precipitation may be interpolated from existing rain gages, and global data sets from remote sensing models to account for remote areas. Precipitation as snow is included. If field data are not available, you can use coarse data from the freely available global data set developed by the Climatic Research Unit (http://www.cru.uea.ac.uk). Within the United States, the PRISM group at Oregon State University provides free precipitation data at a 30-arcsecond resolution. See their website at http://www.prism.oregonstate.edu/ and navigate to ‘800 m Normals’ to download data. 2. Average annual reference evapotranspiration (:math:‘ET_0‘)
Reference evapotranspiration, ET0, is the energy (expressed as a depth of water, e.g. mm) supplied by the sun (and occasionally wind) to vaporize water. Some global products are available on the internet, such as FAO Penman - Monteith method with limited climatic data as described in FAO Irrigation and Drainage Paper 56 using data from the Climatic Research Unit. Reference evapotranspiration depends on elevation, latitude, humidity, and slope aspect. There are countless methodologies, which range in data requirements and precision. If the use of this grid is not possible, develop monthly average grids of precipitation, and maximum and minimum temperatures (http://www.cru.uea.ac.uk), which need to incorporate the effects of elevation when interpolating from observation stations. Data to develop these monthly precipitation and tempera- tures grids follow the same process in the development of the ‘Average Annual Precipitation’ grid, with the added monthly disaggregated grids. A simple way to determine reference Evapotranspiration is the ‘modified Hargreaves’ equation, which generates superior results than the Pennman-Montieth when information is uncertain. The ‘modified Hargreaves’ uses the average of the mean daily maximum and mean daily minimum tem- peratures (Tavg in oC), the difference between mean daily maximum and mean daily minimums (TD), RA is extraterrestrial radiation (RA in MJm−2d−1 and precipitation (P in mm per month), all of which can be relatively easily obtained. Temperature and precipitation data are often available from regional charts or direct measurement. Radiation data, on the other hand, is far more expensive to measure directly but can be reliably estimated from online tools, tables or equations. The reference evapotranspiration could be also calculated monthly and annually using the Hamon equation (Hamon 1961, Wolock and McCabe 1999):
2 PEDHamon = 13.97dD Wt
15.7. Appendix A: Data Sources 261 InVEST User Guide, Release 2.5.6
where d is the number of days in a month, D is the mean monthly hours of daylight calculated for each year (in units of 12 hours), and Wt is a saturated water vapor density term calculated by:
4.95e0.062T W = t 100
where T is the monthly mean temperature in degrees Celsius. Reference evapotranspiration is set to zero when mean monthly temperature is below zero. Then for each year during the time periods analyzed, the monthly calculated PET values at each grid cell are summed to calculate a map of the annual PET for each year. 3. Root restricting layer depth Root restricting layer depth is the soil depth at which root penetration is strongly inhibited be- cause of physical or chemical characteristics. Root restricting layer depth may be obtained from some soil maps. If root restricting layer depth or rootable depth by soil type is not available, soil depth can be used as a proxy. The FAO provides global soil data in their Harmonized World Soil Database: http://www.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML/ Soil data for many parts of the world are also available from the Soil and Terrain Database (SOTER) Programme: http://www.isric.org/projects/soil-and-terrain-database-soter-programme. In the United States free soil data is available from the U.S. Department of Agriculture’s NRCS in the form of two datasets: SSURGO http://soils.usda.gov/survey/geography/ssurgo/ and STATSGO http://soils.usda.gov/survey/geography/statsgo/ . Where available SSURGO data should be used, as it is much more detailed than STATSGO. Where gaps occur in the SSURGO data, STATSGO can be used to fill in the blanks. The root restricting layer depth should be calculated as the maximum depth of all horizons within a soil class component, and then a weighted average of the components should be estimated. This can be a tricky GIS analysis: In the US soil categories, each soil property polygon can contain a number of soil type components with unique properties, and each component may have different soil horizon layers, also with unique properties. Processing requires careful weighting across components and horizons. The Soil Data Viewer (http://soildataviewer.nrcs.usda.gov/), a free ArcMap extension from the NRCS, does this soil data processing for the user and should be used whenever possible. Ultimately, a grid layer must be produced. 4. Plant available water content (PAWC) Plant available water content is a fraction obtained from some standard soil maps. It is defined as the dif- ference between the fraction of volumetric field capacity and permanent wilting point. Often plant avail- able water content is available as a volumetric value (mm). To obtain the fraction divide by soil depth. Soil characteristic layers are estimated by performing a weighted average from all horizons within a soil com- ponent. If PAWC is not available, raster grids obtained from polygon shape files of weight average soil tex- ture (%clay, %sand, %silt) and soil porosity will be needed. See ‘Root Restricting Layer Depth’ above for a description of where to find and how to process soil data. http://hydrolab.arsusda.gov/SPAW/Index.htm has software to help you estimate your PAWC when you have soil texture data. 5. Land use/land cover A key component for all Tier 1 water models is a spatially continuous landuse / land class raster grid. That is, within a watershed, all landuse / land class categories should be defined. Gaps in data that break up the drainage continuity of the watershed will create errors. Unknown data gaps should be approxi- mated. Global land use data is available from the University of Maryland’s Global Land Cover Facility: http://glcf.umiacs.umd.edu/data/landcover/. This data is available in 1 degree, 8km and 1km resolutions. Data for the U.S. for 1992 and 2001 is provided by the EPA in their National Land Cover Data product: http://www.epa.gov/mrlc/.
15.7. Appendix A: Data Sources 262 InVEST User Guide, Release 2.5.6
The simplest categorization of LULCs on the landscape involves delineation by land cover only (e.g., cropland, temperate conifer forest, and prairie). Several global and regional land cover classifications are available (e.g., Anderson et al. 1976), and often detailed land cover classification has been done for the landscape of interest. A slightly more sophisticated LULC classification could involve breaking relevant LULC types into more meaningful categories. For example, agricultural land classes could be broken up into different crop types or forest could be broken up into specific species. The categorization of land use types depends on the model and how much data is available for each of the land types. The user should only break up a land use type if it will provide more accuracy in modeling. For instance, for the water quality model the user should only break up ‘crops’ into different crop types if they have information on the difference in nutrient loading between crops. Along the same lines, the user should only break the forest land type into specific species for the water supply model if information is available on the root depth and evapotranspiration coefficients for the different species. Sample Land Use/Land Cover Table ID Land Use/Land Cover 1 Evergreen Needleleaf Forest 2 Evergreen Broadleaf Forest 3 Deciduous Needleleaf Forest 4 Deciduous Broadleaf Forest 5 Mixed Cover 6 Woodland 7 Wooded Grassland 8 Closed Shrubland 9 Open Shrubland 10 Grassland 11 Cropland (row Crops) 12 Bare Ground 13 Urban and Built-Up 14 Wetland 15 Mixed evergreen 16 Mixed Forest 17 Orchards/Vineyards 18 Pasture 6. Maximum root depth table A valuable review of maximum plant rooting depths is available in Canadell, J., R. B. Jackson, and H. Mooney. 1996, Maximum rooting depth of vegetation types at the global scale. Oecologia 108: 583-595 where 290 observations in the literature are summarized, and it is concluded that rooting depths are more consistent than that previously believed among similar biomes and plant species. The model determines the minimum of root restricting layer depth and rooting depth for an accessible soil profile for water storage. Determinations on how to deal with soil-less systems, such as fractured rock substrates, should be based on expert advice. Effective maximum root depth must be defined for impermeable landuse/land classes, such as urban areas, or water bodies. A rule of thumb is to denote water and urban areas with minimal maximum rooting depths, but a zero value should not be used. The literature values must be converted to mm, and depicted as integer values. Maximum root depths by species and biomes
15.7. Appendix A: Data Sources 263 InVEST User Guide, Release 2.5.6
Root Depth by Species Root Depth by Biome Trees 7.0 m Cropland 2.1 m Shrubs 5.1 m Desert 9.5 m Herbaceous Plants 2.6 m Sclerophyllous Shrubland & Forest 5.2 m Tropical Deciduous Forest 3.7 m Tropical Evergreen Forest 7.3 m Grassland 2.6 m Tropical Grassland/Savanna 15 m Tundra 0.5 m 7. Evapotranspiration coefficient table (:math:‘K_c‘)
Evapotranspiration coefficient (Kc) values for crops are readily available from irrigation and horticulture handbooks. FAO has an online resource for this: http://www.fao.org/docrep/X0490E/x0490e0b.htm. The FAO tables list coefficients by crop growth stage (Kc ini, Kc mid, Kc end), which need to be converted to an annual average Kc because this is an annual water yield model. This requires knowledge about the phenology of the vegetation in the study region (average green-up, die-down dates) and crop growth stages (when annual crops are planted and harvested). Annual average Kc can be estimated as a function of vegetation characteristics and average monthly reference evapotranspiration using the following equation:
P12 K × ET K = m=1 cm om c P12 m=1 ETom
where Kcm is an average crop coefficient of month m (1-12) and ETom is the corresponding refer- ence evapotranspiration. These values can also be calculated using the following spreadsheet: http://ncp- dev.stanford.edu/~dataportal/invest-data/Kc_calculator.xlsx. Values for Kc should be integers between 0-1500. Values for other vegetation can be estimated using Leaf Area Index (LAI) relationships, which is a satellite imagery product derived from NDVI analysis. A typical LAI - Kc relationship might look as follows:
LAI when LAI ≤ 3 K = 3 c 1
Evapotranspiration coefficients need to be applied to non-vegetated class, such as pavement or water bodies. As a rule of thumb, impermeable surfaces and moving water bodies might be given a low Kc value (no zeros should be defined), such as 0.001, to highlight removal of water by drainage. Slow or stagnant water bodies might be given an Kc value of 1. Once evapotranspiration coefficients have been established for all landuse / land classes they must be multiplied by 1000 to obtain the integer value, i.e. Int(Kc x 1000). No zero values are allowed. Sample Evapotranspiration coefficient(:math:‘K_c‘) Table.
15.7. Appendix A: Data Sources 264 InVEST User Guide, Release 2.5.6
ID Vegetation Type Kc 1 Evergreen Needleleaf Forest 1000 2 Evergreen Broadleaf Forest 1000 3 Deciduous Needleleaf Forest 1000 4 Deciduous Broadleaf Forest 1000 5 Mixed Cover 1000 6 Woodland 1000 7 Wooded Grassland 1000 8 Closed Shrubland 398 9 Open Shrubland 398 10 Grassland 650 11 Cropland (Row Crops) 650 12 Bare Ground 1 13 Urban and Built-Up 1 14 Wetland 1200 15 Mixed Evergreen 1000 16 Mixed Forest 1000 17 Orchards/Vineyards 700 18 Pasture 850 19 Sclerophyllous Forests 1000 8. Digital elevation model (DEM) DEM data is available for any area of the world, although at varying resolutions. Free raw global DEM data is available on the internet from NASA - http://asterweb.jpl.nasa.gov/gdem-wist.asp, and USGS - http://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/Elevation_Products and http://hydrosheds.cr.usgs.gov/. Or a final product may be purchased relatively inexpensively at sites such as MapMart (www.mapmart.com). The DEM used in the model must be hydrologically correct meaning that sinks are filled and there are no holes. See the Working with the DEM section of this manual for more information. 1. Consumptive water use The consumptive water use for each land use / land class type should be estimated based on agricultural, forestry, and hydrology literature and/or consultation with local professionals in these fields. The value used in the table is an average for each land use type. For crops, water use can be calculated using information on crop water requirements and scaling up based on area covered by crops. In more general agricultural areas, water use by cattle, agricultural processing, etc. must be considered. For forestry, a similar calculation can be made based on estimates of water use by different forest types. In urban areas, water use may be calculated based on an estimated water use per person and multiplied by the approximate population area per raster cell. Industrial water use must also be considered where applicable. For all of these calculations, it is assumed that the crops, trees, people, etc. are spread evenly across each land use class. 10. Hydropower Watersheds and Sub-watersheds See the Working with the DEM section of this manual for information on generating watersheds and sub-watersheds. The resulting delineation should be checked to ensure that the watersheds accurately represent reality. This reality check may involve talking to a local hydrologist, checking the drainage area for a nearby USGS gage, or doing a back of the envelope calculation for the annual rainfall multiplied by the watershed area and comparing it to the average annual volume of flow into the hydropower station. If you do not have a starting point for sub-watersheds, the global dataset from Hydro1k may be applicable: http://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/gtopo30/hydro. 11. Hydropower Station Information
15.7. Appendix A: Data Sources 265 InVEST User Guide, Release 2.5.6
Detailed information about each hydropower station may only be available from the owner or man- aging entity of the stations. Some information may be available through public sources, and may be accessible online. In particular, if the hydropower plant is located in the United States information may be found on the internet. The first place to check is the National Inventory of Dams (http://crunch.tec.army.mil/nidpublic/webpages/nid.cfm). If a hydropower dam is owned by the Bureau of Reclamation, they should have information on the reservoir on their Dataweb (http://www.usbr.gov/dataweb/). Similar information may be found online at other websites for reser- voirs owned or operated by other government agencies or energy companies. • Calibration: For calibration, data are needed on how much water actually reaches each hydropower station on an average annual basis. Data should be available from the managing entity of the hy- dropower plant. In absence of information available directly from the hydropower operators, data may be available for a stream gage just upstream of the hydropower station. Gages in the U.S. may be managed by the USGS, the state fish and wildlife agency, the state department of ecology or by a local university. The model user should consider whether the gage measures natural or man- aged streamflow and adjust measurements as necessary. The drainage area downstream of the gage and upstream of the hydropower station must also be considered when comparing gaged flow with modeled flow. • Time_period: The design life span of each hydropower station can be obtained from the station owner or operator. Alternative sources may be available online as described above. This value may instead represent the time period of a scenario of interest, which should be equal to or smaller than the life span of the station. • Discount_rate: this rate is defined as how much value the currency loses per year. 12. Seasonality factor (Z) The seasonality factor is used to characterize the seasonality of precipitation in the study area, with possible values ranging from 1 to 10. The values are assigned according to the timing of the majority of rainfall in a year. If rainfall primarily occurs during the winter months, Zhang values should be closer to 10; if most rainfall occurs during the summer months or is more evenly spread out during the year, Zhang values should be closer to 1. Our initial testing efforts of this model in different watersheds in different eco-regions worldwide show that this factor is around 4 in tropical watersheds, 9 in temperate watersheds and 1 in monsoon watersheds.
15.8 Appendix B: Calibration of Water Yield Model
The water yield model is based on a simple water balance where it is assumed that all water in excess of evaporative loss arrives at the outlet of the watershed. The model is an annual average time step simulation tool applied at the pixel level but reported at the sub-basin level. A first run model calibration should be performed using 10 year average input data. For example, if water yield model simulations are being performed under a 1990 land use scenario, climate data (total precipitation and potential evapotranspiration) from 1985 to 1995 should be averaged and used with the 1990 land use map. The other inputs, root restricting layer depth and plant available water content are less susceptible to temporal variability so any available data for these parameters may be used. Observed flow data should be collected from a station furthest downstream in the watershed. As with the climate data, a 10 year average should be used for model calibration. Gauge data is often provided in flow units (i.e m3/s). Since the model calculates water volume, the observed flow data should be converted into units of m3/year. Note, to ensure accuracy, the watershed input being used in the water yield model should have the same approximate area as the contributing watershed area provided with the observed flow data. When assessing the overall accuracy of the model, the mean water yield for the watershed should be compared with the observed depth at the outlet. In nested watersheds or adjacent watersheds, calibration could be carried out on one or two stations (watersheds) and validation of these calibrated watersheds could be carried on the other watershed(s). Before the user starts the calibration process, we highly recommended sensitivity analysis using the observed runoff data. The sensitivity analysis will define the parameters that influence model outputs the most. The calibration can
15.8. Appendix B: Calibration of Water Yield Model 266 InVEST User Guide, Release 2.5.6
then focus on highly sensitive parameters followed by less sensitive ones. As with all models, model uncertainty is inherent and must be considered when analyzing results for decision making. The model is therefore essentially driven more by parameter values (Z, Kc, root depth) then by the individual physical hydrologic processes taking place in the watershed. Since these parameter values are often obtained from literature or experimental studies under varied conditions, a range of values are usually available (see data sources). InVEST Water Yield model uncertainty is best addressed by performing model simulations under maximum, minimum and mean parameter values. Doing so will provide a range of outputs corresponding to plausible actual conditions.
15.9 References
Budyko, M.I. 1974, Climate and Life, Academic, San Diego, California. Donohue, R.J., Roderick, M.L. & McVicar, T.R. 2007, “On the importance of including vegetation dynamics in Budyko’s hydrological model.”, Hydrology and Earth System Sciences, vol. 11, pp. 983-995. Ennaanay, Driss. 2006. Impacts of Land Use Changes on the Hydrologic Regime in the Minnesota River Basin. Ph.D. thesis, graduate School, University of Minnesota. Milly, P.C.D. 1994, “Climate, soil water storage, and the average annual water balance.”, Water Resources Research, vol. 3, no. 7, pp. 2143-2156. Potter, N.J., Zhang, L., Milly, P.C.D., McMahon, T.A. & Jakeman, A.J. 2005, “Effects of rainfall seasonality and soil moisture capacity on mean annual water balance for Australian catchments.”, Water Resources Research, vol. 41. World Commission on Dams (2000). Dams and development: A new framework for decision- making. The Report of the World Commission on Dams. Earthscan Publications LTD, London. Zhang, L., Dawes, W.R. & Walker, G.R. 2001, “Response of mean annual evapotranspiration to vegetation changes at catchment scale. ”, Water Resources Research, vol. 37, pp. 701-708.
15.9. References 267 CHAPTER SIXTEEN
WATER PURIFICATION: NUTRIENT RETENTION
16.1 Summary
Water purification is an essential service provided by ecosystems. InVEST estimates the contribution of vegetation and soil to pu- rifying water through the removal of nutrient pollutants from runoff. The biophysical model uses data on water yield, land use and land cover, nutrient loading and filtration rates and wa- ter quality standards (if they exist) to determine nutrient retention capacity for current and future land use scenarios. The valuation model uses data on water treatment costs and a discount rate to determine the value contributed by the natural system to water purification. It does not address chemical or biological interac- tions besides filtration by terrestrial vegetation (such as in-stream processes) and is less relevant to locations with extensive tile drainage or ditching, strong surface water-ground water interac- tions, or hydrology dominated by infiltration excess (dry regions with flashy rains).
16.2 Introduction
Clean water is a vital service provided by healthy streams, wa- tersheds and river basins. Polluted water is especially harmful to human health. In fact, waterborne illnesses are the leading cause of human disease and death around the world killing more than 3.4 million people annually (World Health Organization). Clean water also provides habitat for aquatic life in streams, rivers and lakes but these habitats require a proper nutrient balance. If nutrients and toxins accumulate in water, fish and other aquatic creatures may be poisoned, along with the people consuming them. Many of these harmful conditions are caused by non-point source pollution, which occurs when a pollution source is distributed over an area or discharged into the atmosphere and incorporated into hydrological flows through rainfall and runoff. There are numerous sources of non-point source pollution, including fertilizer used in agriculture and residential landscaping, and oil that leaks from cars onto roads. When it rains or snows, water flows over the landscape carrying pollutants from these surfaces into streams, rivers, lakes, and the ocean. One way to reduce non-point source pollution is to reduce the amount of pollutants that enter the water body. If this is not possible, ecosystems can provide this service by retaining some non-point source pollutants. For instance, vegetation can remove pollutants by storing them in tissue or releasing them back to the environment in another form.
268 InVEST User Guide, Release 2.5.6
Soils can also store and trap some soluble pollutants. Wetlands can slow flow long enough for pollutants to be taken up by vegetation. Riparian vegetation is particularly important in this regard, often serving as a last defense against pollutants entering a stream. Land use planners from government agencies to environmental groups need information regarding the contribution of ecosystems to mitigating water pollution. Specifically, they require information pertaining to the value of every part of a watershed for maintaining water quality so that conservation may be targeted to the areas most important for protecting a safe water supply for people and aquatic life. They can also use this information to avoid impacts in areas that currently contribute the most to filtering out pollutants. The InVEST Tier 1 model provides this information for non-point source pollutants. We have designed the model to deal with nutrient pollutants (nitrogen and phosphorous), but the model can be used for other kinds of contaminants (persistent organics, pathogens etc.) if data are available on the loading rates and filtration rates of the pollutant of interest.
16.3 The Model
The InVEST Water Purification Nutrient Retention model calculates the amount of nutrient retained on every pixel then sums and averages nutrient export and retention per sub-watershed. The pixel-scale calculations allow us to represent the heterogeneity of key driving factors in water yield such as soil type, precipitation, vegetation type, etc. However, the theory we are using as the foundation of this set of models was developed at the sub-watershed to watershed scale. We are only confident in the interpretation of these models at the sub-watershed scale, so all outputs are summed and/or averaged to the sub-basin scale. We do continue to provide pixel-scale representations of some outputs for calibration and model-checking purposes only. These pixel-scale maps are not to be interpreted for understanding of hydrological processes or to inform decision making of any kind. InVEST also calculates the economic value that nutrient retention provides through avoided treatment costs. It inte- grates data on the magnitude of overland flow, pollutant loading, the capacity of different vegetation types to filter pollutants, the cost of water treatment (for pollutants of interest), and feasibility to meet water quality standards. The model’s limitations are that it runs on an annual average basis, can only assess one pollutant per run, does not address chemical or biological interactions besides filtration by terrestrial vegetation, and it may provide an inaccurate marginal cost for pollutant removal when pollutant costs relative to pollutant concentration are non-linear. The model assumes that non-point sources of water pollution result from export that can be mitigated by vegetation serving as intercepting filters. It also assumes that water flows downslope along natural flowpaths, so it may be less relevant in areas with tile drainage and extensive ditching practices. It does not consider the role of ecosystems in affecting point-source pollutants. It also may be less relevant where there is significant groundwater surface water interaction and in dry eco-regions.
16.3.1 How it Works
The model runs on a gridded map. It estimates the quantity and value of pollutants retained for water purification from a landscape in three components. The first step calculates annual average runoff from each parcel. See the Hydropower chapter for information on water yield. In the second step, we determine the quantity of pollutant retained by each parcel on the landscape. First, we estimate how much pollutant is exported from each parcel, based on export coefficients the user inputs. Export coefficients, de- veloped by Reckhow et al. 1980, are annual averages of pollutant fluxes derived from various field studies that measure export from parcels within the United States. Since these coefficients are average fluxes, we include a hydrological sensitivity score that accounts for differences in condition between the fields where the measures were developed and the conditions where the user is applying the model. We do this with the following equation:
ALVx = HSSx · polx
16.3. The Model 269 InVEST User Guide, Release 2.5.6
where ALVx is the Adjusted Loading Value at pixel x, polx is the export coefficient at pixel x, and HSSx is the Hydrologic Sensitivity Score at pixel x which is calculated as:
λx HSSx = λW
where λx is the runoff index at pixel x, calculated using the following equation, and λW is the mean runoff index in the watershed of interest. ! X λa = log Yu U
P where U Yu is the sum of the water yield of pixels along the flow path above pixel x (it also includes the water yield of pixel x). Once we know how much pollutant leaves each pixel, we can determine how much of that load is retained by each downstream pixel, as surface runoff moves the pollutant toward the stream. The model routes water down flow paths determined by slope, and allows each pixel downstream from a polluting pixel to retain pollutant based on its land cover type and that land cover type’s ability to retain the modeled pollutant. We do not account for saturation of uptake. By following the pollutant load of each pixel all the way downstream to a water body, the model also tracks how much pollutant reaches the stream. The table below describes how this removal from routing and hydraulic connectivity is done:
The model then aggregates the loading that reaches the stream from each pixel to the sub-watershed then to the watershed level. The user can then compare this load (adding the point sources loadings if any) to a known (observed or simulated using another water quality model) measurement and adjust export coefficients and removal efficiencies (vegetation retention) as needed until the modeled load matches the measured load for each point of interest. The user should consider the likely impact of in-stream processes in any calibration work as this model does not include in-stream processes. pixel To calculate the amount of service delivered, the model decreases retention by the amount of ‘allowed’ pollution in the water body of interest, if an allowed amount is given. This step accounts for regulations that define a concentration of contaminants of concern. In other words, in water bodies where there is a water quality standard, watershed retention of nutrients that would lead to river concentrations below that standard should not be counted as an environmental service since people in effect do not care if that low amount of pollution occurs. In that sense, the model does not give credit to retention of nutrients below the user-defined threshold. If a threshold is given, the service level is calculated
16.3. The Model 270 InVEST User Guide, Release 2.5.6 in biophysical terms as follows: thresh net = retained − x x contrib
where retainedx is the amount of retention calculated as in the table above, thresh is the total allowed annual load for the pollutant of interest (thresh_p for phosphorous, thresh_n for nitrogen) and contrib is the number of pixels on the landscape. Pixel values are then summed (nret_sm) or averaged (nret_mn) to the sub-watershed scale to give sub-watershed service outputs in biophysical terms. Once the service level (nret) is determined, we can (optionally) calculate the value of this service provided by each sub-watershed based on the avoided treatment costs that retention by natural vegetation and soil provides. We make this calculation as follows: T −1 X 1 wp_V alue = Cost(p) ∗ retained ∗ x x (1 + r)t t=0
Where:
wp_V aluex is the value of retention for sub-watershed x. Cost(p) is the annual treatment cost in $(currency)/kg for the pollutant of interest (p).
retainedx is the total pollutant retained by sub-watershed x T is the time span being considered for the net present value of water treatment r is the discount rate used for calculating net present value The sub-watershed values are then summed to the watershed to determine the water purification value per watershed.
16.3.2 Limitations and Simplifications
The model has a number of assumptions. First, since the model was developed for watersheds and landscapes domi- nated by saturation excess runoff hydrology, it may be less applicable to locations where the hydrology is determined by rainfall intensity; in areas where flashy rains are predominant and where infiltration excess runoff occurs. This kind of runoff is the result of intense rains that saturate only the top soil layer, not the entire profile. However, the model’s use of a runoff index and hydraulic routing should sufficiently adjust for this. Second, the model can only assess one pollutant per run. If the user wishes to model several pollutants, but does not have data on loadings and filtration rates for each pollutant, choose a pollutant that acts as a surrogate in predicting loadings for other pollutants. The most common surrogate is phosphorus because heavy phosphorus loadings are often associated with other pollutants such as nitrogen, bacteria and suspended solids. However, using a pollutant surrogate should be approached with caution. Alternatively, the user can run the model multiple times using export values and retention coefficients for each pollutant. In general, the model can only assess pollutants that are susceptible to export via surface and subsurface flows. Third, the model does not address any chemical or biological interactions that may occur from the point of loading to the point of interest besides filtration by terrestrial vegetation. In reality, pollutants may degrade over time and distance through interactions with the air, water, other pollutants, bacteria or other actors. Fourth, the model assumes that there is continuity in the hydraulic flow path. The user should be aware of any discontinuity in the flow path. Tile drainage and ditches could create short cuts for pollutant movement and run pollutant directly to streams. Finally, in some cases the model may provide an inaccurate marginal cost for pollutant removal. The full marginal cost of removing a unit volume of pollutants is difficult to estimate due to the complexity of the treatment process. The marginal cost may not be a constant value but instead a function of decreasing cost per additional unit volume of pollutant as the total volume increases. Also, the cost of treatment may change over time as technology improves or water quality standards evolve.
16.3. The Model 271 InVEST User Guide, Release 2.5.6
16.4 Data Needs
Here we outline the specific data used by the model. See the appendix for detailed information on data sources and pre-processing. For all raster inputs, the projection used should be defined, and the projection’s linear units should be in meters. 1. Digital elevation model (DEM) (required). A GIS raster dataset, with an elevation value for each cell. Make sure the DEM is corrected by filling in sinks, and if necessary ‘burning’ hydrographic features into the elevation model (recommended when you see unusual streams). To ensure proper flow routing, the DEM should extend beyond the watersheds of interest, rather than being clipped to the watershed edge. See the Working with the DEM section of this manual for more information. Name: File can be named anything, but avoid spaces in the name and less than 13 characters Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with an elevation value for each cell given in meters above sea level. Sample data set: \InVEST\Base_Data\Freshwater\dem 2. Soil depth (required). A GIS raster dataset with an average soil depth value for each cell. The soil depth values should be in millimeters . Name: File name can be anything, but avoid spaces in the name and less than 13 characters. Format: Standard GIS raster file, with an average soil depth in millimeters for each cell. Sample data set: \InVEST\Base_Data\Freshwater\soil_depth 3. Precipitation (required). A GIS raster dataset with a non-zero value for average annual precipitation for each cell. The precipitation values should be in millimeters. Name: File can be named anything, but avoid spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with precipitation values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\precip 4. Plant Available Water Content (required). A GIS raster dataset with a plant available water content value for each cell. Plant Available Water Content fraction (PAWC) is the fraction of water that can be stored in the soil profile that is available for plants’ use. Name: File can be named anything, but avoid spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with available water content values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\pawc 5. Average Annual Potential Evapotranspiration (required). A GIS raster dataset, with an annual average evapotranspiration value for each cell. Potential evapotranspiration is the potential loss of water from soil by both evaporation from the soil and transpiration by healthy Alfalfa (or grass) if sufficient water is available. The evapotranspiration values should be in millimeters. Name: File can be named anything, but avoid spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with potential evapotranspiration values for each cell. Sample data set: \InVEST\Base_Data\Freshwater\eto 6. Land use/land cover (required). A GIS raster dataset, with an LULC code for each cell. The LULC code should be an integer.
16.4. Data Needs 272 InVEST User Guide, Release 2.5.6
Name: File can be named anything, but avoid spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with an integer LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.). These codes must match LULC codes in the Biophysical Table. Sample data set: \InVEST\Base_Data\Freshwater\landuse_90 7. Watersheds (required). A shapefile of polygons. This is a layer of watersheds such that each watershed contributes to a point of interest where water quality will be analyzed. See the Working with the DEM section for information on creating watersheds. Name: File can be named anything, but avoid spaces. Format: Standard GIS shapefile , with unique integer values for each watershed in the ws_id field Sample data set: \InVEST\Base_Data\Freshwater\watersheds.shp 8. Sub-watersheds (required). A shapefile of polygons. This is a layer of sub-watersheds, contained within the Watersheds (described above) which contribute to the points of interest where water quality will be analyzed. See the Working with the DEM section for information on creating sub-watersheds. Due to limitations in ArcMap geoprocessing, the maximum size of a sub-watershed that can be used in the Water Purification model is approximately the equivalent of 4000x4000 cells, with cell size equal to the smallest cell size of your input layers. Name: File can be named anything, but avoid spaces. Format: A shapefile of polygons with unique integers for each sub-watershed in the subws_id field. Sample data set: \InVEST\Base_Data\Freshwater\subwatersheds.shp 9. Biophysical Table (required). A table of land use/land cover (LULC) classes, containing data on water quality coefficients used in this tool. NOTE: these data are attributes of each LULC class rather than attributes of individual cells in the raster map. Name: File can be named anything. File type: *.dbf or *.mdb Rows: Each row is an LULC class. Columns: Each column contains a different attribute of each land use/land cover class, and must be named as follows: 1. lucode (Land use code): Unique integer for each LULC class (e.g., 1 for forest, 3 for grassland, etc.), must match the LULC raster above. 2. LULC_desc: Descriptive name of land use/land cover class (optional) 3. root_depth: The maximum root depth for vegetated land use classes, given in integer millimeters. Non-vegetated LULCs should be given a value of 1. 4. etk: The evapotranspiration coefficient for each LULC class, used to obtain actual evapotranspiration by using plant energy/transpiration characteristics to modify the reference evapotranspiration, which is based on alfalfa (or grass). Coefficients should be multiplied by 1000, so that the final etk values given in the table are integers ranging between 1 and 1500 (some crops evapotranspire more than alfalfa in some very wet tropical regions and where water is always available). 3. load_n / load_p: The nutrient loading for each land use. If nitrogen is being evaluated, supply values in load_n, for phosphorus, supply values in load_p. The potential for terrestrial loading of water quality impairing constituents is based on nutrient export coefficients. The nutrient loading values are given as integer values and have units of g. Ha-1 yr -1.
16.4. Data Needs 273 InVEST User Guide, Release 2.5.6
4. eff_n / eff_p: The vegetation filtering value per pixel size for each LULC class, as an integer percent between zero and 100. If nitrogen is being evaluated, supply values in eff_n, for phosphorus, supply values in eff_p. This field identifies the capacity of vegetation to retain nutrient, as a percentage of the amount of nutrient flowing into a cell from upslope. For example if the user has data describing that wetland of 5000 m2 retains 82% of nitrogen, then the retention efficiency that the he should input into this field for eff_n is equal to (82/5000 * (cell size)2). In the simplest case, when data for each LULC type are not available, high values (60 to 80) may be assigned to all natural vegetation types (such as forests, natural pastures, wetlands, or prairie), indicating that 60-80% of nutrient is retained. An intermediary value also may be assigned to features such as contour buffers. All LULC classes that have no filtering capacity, such as pavement, can be assigned a value of zero. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Biophysical_Models Example : Case with 6 LULC categories, where potential evapotranspiration, root depth and nutrient (both N and P) filtration efficiencies do not vary among LULC categories, while nutrient loadings do. LULC_desc lucode etk root_depth load_n eff_n load_p eff_p Low Density Residential 1 1 1 7000 0 1000 0 Mid Density Residential 2 1 1 7250 0 1100 0 High Density Residential 3 1 1 7500 0 1200 0 Very High Density Residential 4 1 1 7750 0 1300 0 Vacant 5 1 1 4000 0 100 0 Commercial 6 1 1 13800 0 3000 0 9. Threshold flow accumulation value (required). Integer value defining the number of upstream cells that must flow into a cell before it’s considered part of a stream. This is used to generate a stream layer from the DEM. The default is 1000. If the user has a map of stream lines in the watershed of interest, he/she should compare it with the Outpuv_stream map that is output by the model. This value also needs to be well estimated in watersheds where tile drainage and ditches are present. This threshold expresses where hydraulic routing is discontinued and where retention stops and the remaining pollutant will be exported to the stream. 10. Water Purification Valuation table. This is a table containing valuation information for each of the points of interest. There must be one row for each watershed in the Watersheds layer. Name: File can be named anything. File type: *.dbf or *.mdb for ArcGIS models, the standalone model requires a .csv file Rows: Each row corresponds to a watershed. Columns: Each column contains a different attribute of each watershed and must be named as follows: 1. ws_id (watershed ID): Unique integer value for each watershed, which must correspond to values in the Watersheds layer. 2. cost: Annual cost of nutrient removal treatment in $ / kg removed. Floating point value. 3. time_span: Number of years for which net present value will be calculated. Integer value. This could be the time span (number of years) of either the same LULC scenario or the water treatment plant life span. 4. discount: The rate of discount over the time span, used in net present value calculations. Floating point value. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Water_Purification_Valuation Example for 3 watersheds/points of interest, from the sample data set: ws_id cost time_span discount 0 24 15 5 1 24 25 5 2 24 15 5
16.4. Data Needs 274 InVEST User Guide, Release 2.5.6
11. Water Purification threshold table. A table containing annual nutrient load threshold information for each of the points of interest. There must be one row for each watershed in the Watersheds layer. Name: File can be named anything. File type: *.dbf or *.mdb for ArcGIS models, the standalone model requires a .csv file Rows: Each row corresponds to a watershed. Columns: Each column contains a different attribute of each watershed and must be named as follows: 1. ws_id (watershed ID): Unique integer value for each watershed, which must correspond to values in the Watersheds layer. 2. thresh_n / thresh_p: The total critical annual nutrient loading allowed for the nutrient of interest at the point of interest. Floating point value. It has units of Kg.yr-1. Sample data set: C:\Invest\Base_Data\Freshwater\Water_Tables.mdb\Water_Purification_Threshold
16.5 Running the Model
Before running the Water Purification Nutrient Retention model, make sure that the InVEST toolbox has been added to your ArcMap document, as described in the Getting Started chapter of this guide. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. • Create a workspace on your computer hard drive if you are using your data. The pathname to the workspace should not have spaces. All your output files will be saved here. For simplicity, you could create a folder in your workspace called “Input” and place all your input files here. It is not necessary to place input files in the workspace, but this will make it easier to view the data you use to run your model. If this is your first time using InVEST and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you installed the InVEST files on your C drive (as described in the Getting Started chapter), you should see a folder named /InVEST/WP_Nutrient_Retention. This folder will be your workspace. The input files are in /InVEST/Base_Data/Freshwater/. • Open an ArcMap document to run the model. • Locate the InVEST toolbox in ArcToolbox. ArcToolbox should be open in ArcMap, but if it is not, click on the ArcToolbox symbol. See the Getting Started chapter if you do not see the InVEST toolbox. • Click the plus sign on the left side of the InVEST toolbox to expand the list of tools. Double-click on Nutri- ent_Retention. Three options will appear: Water Yield, Nutrient Removal, and Valuation. Water Yield must be run first, Nutrient Removal second, and Valuation last. The scripts MUST be run in this order because the output from a previous script is required for the next script. • Click on Water Yield. • An interface will appear like the one above that indicates default (sample data) file names, but you can use the file buttons or drop-down arrows to browse to your data. When you place your cursor in each space, you can read a brief description of the data requirements in the right side of the interface. Refer to the Data Needs section for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires data. • After entering all required data, click OK. The script will run, and its progress will be indicated by a “Progress dialogue”. • Load the output files into ArcMap using the ADD DATA button. • To view the attribute data of output files, right click a layer and select OPEN ATTRIBUTE TABLE.
16.5. Running the Model 275 InVEST User Guide, Release 2.5.6
• Now you are ready to run Nutrient Removal. Follow the same steps as for Water Yield. Note that an output from Water Yield, Output\Pixel\wyield, is a required input to Nutrient Retention. Make sure to select one of the Nutrient Type boxes, the model needs one of the two to be checked to run You may see (optional) after Nitrogen or Phosphorus, but you still need to check the box of the nutrient you are interested in. The interface is below: • When the script completes running, its results will be saved in the Output and Service folders. • Load the output files into ArcMap using the ADD DATA button. • Finally, you have the option to run Valuation. Three outputs from Nutrient Removal are required, Ser- vice\nret_sm, Output\nutrient_watershed.dbf, and Output\nutrient_ subwatershed.dbf. The interface is below: • When the script completes running, its results will be saved in the Service folder. • Load the output files into ArcMap using the ADD DATA button. • To view the attribute data of output files, right click a layer and select OPEN ATTRIBUTE TABLE.
16.6 Interpreting Results
The following is a short description of each of the outputs from the Water Purification model. Final results are found in the Output and Service folders within the Workspace specified for this model. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix. • Output\adjl_mn (kg/ha): Mean adjusted load per sub-watershed.
16.6. Interpreting Results 276 InVEST User Guide, Release 2.5.6
16.6. Interpreting Results 277 InVEST User Guide, Release 2.5.6
• Output\adjl_sm (kg/sub-watershed, not /ha): Total adjusted load per sub-watershed. • Service\nret_sm (kg/sub-watershed, not /ha): Total amount of nutrient retained by each sub-watershed. • Service\nret_mn (kg/ha): Mean amount of nutrient retained by each sub-watershed. • Output\nexp_mn (kg/ha): Mean amount of nutrient per sub-watershed that is exported to the stream. • Output\nexp_sm (kg/sub-watershed, not /ha): Total amount of nutrient per sub-watershed that is exported to the stream. • Output\nutrient_subwatershed.dbf: Table containing biophysical values per sub-watershed, with fields as follows: – nut_export (kg/sub-watershed, not /ha): Total amount of nutrient exported to the stream per sub-watershed. – nut_retain (kg/sub-watershed, not /ha): Total amount of nutrient retained by the landscape in each sub- watershed. • Output\nutrient_watershed.dbf: Table containing biophysical values per watershed, with fields as follows: – nut_export (kg/watershed, not /ha): Total amount of nutrient exported to the stream per watershed. – nut_retain (kg/watershed, not /ha): Total amount of nutrient retained by the landscape in each watershed. • Service\nut_val (currency/timespan): The economic benefit per sub-watershed of filtration by vegetation deliv- ered at the downstream point(s) of interest over the specified timespan. THIS OUTPUT REPRESENTS THE ENVIRONMENTAL SERVICE OF WATER PURIFICATION IN ECONOMIC TERMS. It may be useful for identifying areas where investments in protecting this environmental service will provide the greatest returns. Variation in this output with scenario analyses (by running and comparing different LULC scenarios) will indi- cate where land use changes may have the greatest impacts on service provision. • Service\nutrient_value_subwatershed.dbf: Table containing economic values per sub-watershed, with fields as follows: – nut_export/nut_retain: Same as for nutrient_subwatershed.dbf. – nut_value (currency/timespan): Value of the sub-watershed landscape for retaining nutrient over the spec- ified timespan. • Service\nutrient_value_watershed.dbf: Table containing economic values per watershed, with fields as fol- lows: – nut_export/nut_retain: Same as for nutrient_watershed.dbf. – nut_value (currency/timespan): Value of the watershed landscape for retaining nutrient over the specified timespan. These outputs provide an interim insight into the dynamics of pollutant loading, transport and filtration in a watershed. The model will be most informative if it is used in collaboration with experts in hydrology familiar with the watershed. In case model coefficients require adjustment and to guard against erroneous data input, it is recommended that model outputs are verified with field data mimicking pollutant loading and watershed transport processes.
16.7 Appendix: Data Sources
This is a rough compilation of data sources and suggestions about finding, compiling, and formatting data. This section should be used for ideas and suggestions only. It will be updated as new data sources and methods become available. In general, the FAO Geonetwork could be a valuable data source for different GIS layers for users outside the United States: http://www.fao.org/geonetwork/srv/en/main.home. 1. Digital elevation model (DEM)
16.7. Appendix: Data Sources 278 InVEST User Guide, Release 2.5.6
DEM data is available for any area of the world, although at varying resolutions. Free raw global DEM data is available on the internet from the World Wildlife Fund - http://www.worldwildlife.org/freshwater/hydrosheds.cfm. NASA provides free global 30m DEM data at http://asterweb.jpl.nasa.gov/gdem-wist.asp. As does USGS - http://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/Elevation_Products and http://hydrosheds.cr.usgs.gov/. Or, it may be purchased relatively inexpensively at sites such as MapMart (www.mapmart.com). The hydrological aspects of the DEM used in the model must be correct. Please see the Working with the DEM section of this manual for more information. 2. Soil depth Soil depth may be obtained from standard soil maps. Coarse, yet free global soil characteris- tic data are available at http://www.ngdc.noaa.gov/seg/cdroms/reynolds/reynolds/reynolds.htm. The FAO also provides global soil data in their Harmonized World Soil Database: http://www.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML/. In the United States free soil data is available from the U.S. Department of Agriculture’s NRCS in the form of two datasets: SSURGO http://soils.usda.gov/survey/geography/ssurgo/ and STATSGO http://soils.usda.gov/survey/geography/statsgo/ . Where available SSURGO data should be used, as it is much more detailed than STATSGO. Where gaps occur in the SSURGO data, STATSGO can be used to fill in the blanks. Soil depth should be calculated as the maximum depth of all horizons within a soil class component, and then a weighted average of the components should be estimated. This can be a tricky GIS analysis: In the US soil categories, each soil property polygon can contain a number of soil type components with unique properties, and each component may have different soil horizon layers, also with unique prop- erties. Processing requires careful weighting across components and horizons. The Soil Data Viewer (http://soildataviewer.nrcs.usda.gov/), a free ArcMap extension from the NRCS, does this soil data pro- cessing for the user and should be used whenever possible. Ultimately, a grid layer must be produced. Data gaps, such as urban areas or water bodies need to be given appropriate values. Urban areas and water bodies can be thought of having zero soil depth. 3. Land use and land cover A key component for all water models is a spatially continuous land use and land cover raster grid. That is, within a watershed, all land use and land cover categories should be defined. Gaps in data that break up the drainage continuity of the watershed will create errors. Unknown data gaps should be approximated. The more detailed and descriptive these files are the better accuracy and modeling results. Global land cover data is available from the University of Maryland’s Global Land Cover Facility: http://glcf.umiacs.umd.edu/data/landcover/. This data is available in 1 degree, 8km and 1km resolutions. Data for the U.S. for 1992 and 2001 is provided by the EPA in their National Land Cover Data product: http://www.epa.gov/mrlc/. The simplest categorization of LULCs on the landscape involves delineation by land cover only (e.g., cropland, temperate conifer forest, prairie). Several global and regional land cover classifications are available (e.g., Anderson et al. 1976), and often detailed land cover classification has been done for the landscape of interest. A slightly more sophisticated LULC classification could involve breaking relevant LULC types into more meaningful types. For example, agricultural land classes could be broken up into different crop types or forest could be broken up into specific species. The categorization of land use types depends on the model and how much data is available for each of the land types. The user should only break up a land use type if it will provide more accuracy in modeling. For instance, for the Water Purification: Nutrient Retention model the user should only break up ‘crops’ into different crop types if they have information on the difference in nutrient loading between crops. Along
16.7. Appendix: Data Sources 279 InVEST User Guide, Release 2.5.6
the same lines, the user should only break the forest land type into specific species for the water supply model if information is available on the root depth and evapotranspiration coefficients for the different species. 4. Watersheds / Sub-watersheds Watersheds should be delineated by the user, based on the location of reservoirs or other points of interest. Exact locations of specific structures, such as reservoirs, should be obtained from the managing entity or may be obtained on the web at sites such as the National Inventory of Dams (http://crunch.tec.army.mil/nidpublic/webpages/nid.cfm). Watersheds that contribute to the points of interest must be generated. If known correct watershed maps exist, they should be used. Otherwise, watersheds and sub-watersheds can be generated in ArcMap using a hydrologically-correct digital elevation model. Due to limitations in ArcMap geoprocessing, the max- imum size of a sub-watershed that can be processed by the Nutrient Retention tool is approximately the equivalent of 4000x4000 cells, at the smallest cell size of all input grids. See the Working with the DEM section of this manual for more information on generating watersheds and sub-watersheds. 5. Nutrient Loading Coefficients Examples of export and loading coefficients can be found in the EPA PLOAD User’s Manual http://www.epa.gov/waterscience/basins/b3docs/PLOAD_v3.pdf and in the Wetlands Regulatory Assis- tance Program publication http://el.erdc.usace.army.mil/elpubs/pdf/tnwrap04-3.pdf. Note that the exam- ples in the EPA guide are in lbs/ac/yr and would need to be converted to kg/ha/yr. Phosphorus is a common water quality proxy because it incorporates both dissolved and particulate nutri- ent loadings, is well associated with surface runoff, and is usually the limiting nutrient for fresh water sys- tems. The table below shows default phosphorus export coefficients largely based on values from USEPA manuals, and research studies in the US. The bottom three rows are used solely for direct untreated waste water discharge (i.e. untreated sewage piped into water systems) from urban areas commonly found in developing countries. If local data / approximations for Phosphorus export coefficients exist they can be used to replace default values in the table. Table : Example Phosphorus and Nitrogen export coefficients (Reckhow et al. 1980) Landuse Nitrogen Export Coefficient (kg/ha/yr) Phosphorus Export Coefficient (kg/ha/yr) Forest 1.8 0.011 Corn 11.1 2 Cotton 10 4.3 Soybeans 12.5 4.6 Small Grain 5.3 1.5 Pasture 3.1 0.1 Feedlot or Dairy 2900 220 Idle 3.4 0.1 Residential 7.5 1.2 Business 13.8 3 Industrial 4.4 3.8 The loading proxy may also aggregate several indicators, agreed upon between managers, such as an algorithm that aggregates phosphorus, nitrates, and other constituents. Alternatively, a manager may begin using values from EPA table as a starting point to generate discussion, and then alter values based on local expert opinion and stakeholder feedback. 6. Removal Efficiencies (*eff_n*, *eff_p*) These values are used to incorporate the effects of natural vegetation that buffer potential water quality impairment downhill from sources. To develop these values, all land class pixels that contain natural vegetation (such as forests, natural pastures, wetlands, or prairie) are assigned high values and vegetation
16.7. Appendix: Data Sources 280 InVEST User Guide, Release 2.5.6
that has no or little filtering value receives a value of zero. All values should fall between 0 and 100. Consult with a hydrologist if you’re not certain about the assignment of specific values. 7. Calibration Data (calib) Calibration data is needed for ensuring that the Tier 1 Water Purification: Nutrient Retention model results match well with reality. Most often calibration data may be obtained from water quality monitoring that is already in place. If the point of interest is a water supply intake, the drinking water entity will most likely collect water quality at the point of intake. If the point of interest is in a stream or lake, the water quality may have been tested by a public agency. Most likely if the location is of interest in terms of meeting a water quality standard, data should be available. In the U.S. the user may contact or look up online their state environmental agency, EPA, fish and wildlife service, or any local universities conducting research on the water body. Once data is collected, the user may have to convert the values into actual pollutant loads and/or correlate a measured pollutant with a proxy modeled pollutant. In addition to correlation analysis, other calibration methods such as Nash Coefficient, ranking analysis, and graphical comparison could be used. 8. Critical Annual Load (*Ann_Load*) Gathering information on water quality standards or targets should be part of the formulation of modeling objectives. If the target to be met is a drinking water target, standards may be set by the federal, state or local level (whichever standard is the most stringent). The table below provides some general drinking water standards set by global and national agencies. Selected Drinking Water Standards by World Health Organization, European Union, and US EPA. (Ash- bolt et al. 2001)
These standards are set for point of use, meaning that the standard at the point of interest, where water supply will be drawn, may be more relaxed than these standards if water treatment is in place. In-situ water quality standards (for rivers, lakes and streams) may also be set at the national, state and local level. They may be the same across all water bodies of the same type (in rivers, for example) or they may vary depending on the established use of the water body or the presence of endangered species. In the U.S. Total Maximum Daily Loads of various pollutants are typically established by state regulatory agencies in compliance with the Clean Water Act. States report information on TMDLs to the U.S. EPA on specific waterways http://www2.ctic.purdue.edu/kyw/tmdl/statetmdllists.html. 9. Marginal pollutant removal costs (cost) The cost to remove pollutants may vary greatly for each point of interest. If the point of interest is a water supply outtake, this value should be obtained from the water treatment entity who uses and treats the water. Calculations may need to be performed to transform actual costs to cost per unit volume of pollutant, and correlations may need to be run between a proxy pollutant and other pollutants that the treatment process removes. If a more general cost of treatment is sought, the user may consult engineering texts or literature to obtain average costs. The user must be sure to bring these costs into present value and make adjustments as necessary depending on the location and type of treatment.
16.7. Appendix: Data Sources 281 InVEST User Guide, Release 2.5.6
If the point of interest is an in situ water quality target, the marginal pollutant removal cost is much more difficult to obtain. The user may be able to estimate the cost of an additional unit volume of pollutant in terms of fish populations, lost revenue for recreation, or a fine, but this may be a complicated calculation not worth the effort at this level of modeling. The user may choose to assign a cost of one to save time while still obtaining relative results useful in comparing scenarios.
16.8 References
Anderson, J. R., et al. 1976. A Land Use And Land Cover Classification System For Use with Remote Sensor Data: Geological Survey Professional Paper 964. Edited by NJDEP, OIRM, BGIA, 1998, 2000, 2001, 2002, 2005. Ashbolt, N.J., Grabow, W.O.K. and Snozzi, M. 2001. “Indicators of microbial water quality,” in Water Quality: Guidelines, Standards and Health, L. Fretwell and J. Bartram, Editors. 2001, World Health Organization (WHO), IWA Publishing: London, U.K. Reckhow, K.H., Beaulac, M.N. & Simpson, J.T. 1980, Modeling Phosphorus loading and lake response under uncer- tainty: A manual and compilation of export coefficients., U.S. Environmental Protection Agency, Washington, D.C. Uusi Kamppa, J., E. Turtola, H. Hartikainen, T. Ylaranta. 1997. The interactions of buffer zones and phosphorous runoff. In Buffer zones: Their processes and potential in water protection, eds. 14. Haycock, T. Burt, K. Goulding, and G. Pinay, 43–53. Hertfordshire, UK: Quest Environmental.
16.8. References 282 CHAPTER SEVENTEEN
SEDIMENT RETENTION MODEL: AVOIDED DREDGING AND WATER QUALITY REGULATION
17.1 Summary
Reservoirs are linked to a number of environ- mental services, including the generation of en- ergy through reservoir hydropower production, irrigation of crops and recreational activities. Erosion and sedimentation of watersheds can lead to decreased hydropower output, structural damage to reservoirs and other water infrastruc- ture, and flooding. InVEST estimates the ca- pacity of a land parcel to retain sediment us- ing data on geomorphology, climate, vegetation and management practices. These estimates are combined with data on sediment removal costs, reservoir design, and a discount rate to calculate the avoided cost of sediment removal. Limitations of the model include negligence of mass erosion, inadequate information about sediment removal costs, and simplified LULC clas- sifications.
17.2 Sediment Updates to InVEST 3.0
As part of a continuing effort to improve our models we are developing the next generation open source platform of InVEST. The sedimentation model has a new version of this model available in versions of InVEST 2.5.0 and later. It can be found in the Windows start menu under All Programs -> InVEST [version] -> Freshwater -> Sedimentation. Differences in the 3.0 version of this model from the original ArcGIS version will be indicated below. A summary of the changes are below: • Improved runtime performance, stability, and error messages during a runtime failure. • C and P values should be stored in their original floating point state (the ArcGIS version requires the values to be multiplied by 1000 and stored as integers in the biophysical table. • The 3.0 flow algorithm uses a D-infinity flow whereas the ArcGIS version used D8. • The 3.0 sediment model uses a modern LS factor for two dimensional surfaces from Desmet and Govers (1996):
283 InVEST User Guide, Release 2.5.6
2m+1 m+1 Ai,j−in + D − Ai,j−in Li,j = m+2 m m D · xi,j · (22.13)
17.3 Introduction
Erosion and sedimentation are natural processes that contribute to healthy ecosystems, but too much may have severe consequences. Excessive erosion can reduce agricultural productivity, increase flooding and pollutant transport, and threaten bridges, railroads and power infrastructures. Erosion can lead to sediment build-up, which strains water infrastructures, such as reservoirs and flood control systems, and increases water treatment costs. Sedimentation is particularly problematic for reservoirs, which are designed to retain sediment as water is released. Regular sediment removal can avoid some of these issues but this involves expensive maintenance costs. The magnitude of sediment transport in a watershed is determined by several factors. Natural variation in soil proper- ties, precipitation patterns, and slope create patterns of erosion and sediment runoff. Vegetation holds soil in place and captures sediment moving overland. However, changes in land management practices can alter the sediment retention capacity of land by removing important vegetation. There are many clear examples of the effects of LULC change on erosion and sedimentation. Forest fires that clear significant areas of vegetation are often followed by mudslides when heavy rains occur (Meyer et al. 2001). After the fire the vegetation that once held sediment in place no longer exists and the top layers of soil can be carried downstream by overland runoff. Deforestation results in a similar process, although in some cases it may occur on longer time scales. Even in areas where land cover remains the same, a change in land use practice can alter the sediment retention capacity of the landscape. For example, moving from no-till to till agriculture has been shown to increase the rate of soil erosion. The continuous accumulation of increased sediment loads as a result of changes in LULC can cause serious problems such as increasing siltation rate, and increasing dredging costs that were not anticipated during the original design of reservoir infrastructure, maintenance and operation plans. To reduce the damages and costs associated with sedimentation, land, water and reservoir managers require information regarding the extent to which different parts of a landscape contribute to sediment retention, and how land use changes may affect this retention. Such information can support decisions by government agencies, businesses, and NGOs. For example, a power company operating a hydropower reservoir may elect to conserve upstream forests that maintain a sediment retention service if the cost of conserving the forests is less than the costs of reduced hydropower potential, sediment removal, and dam replacement. Maps showing which forest parcels offer the greatest sediment retention benefits would help the power company maximize returns on their investment. InVEST aims to provide these kinds of information. The outputs from these models will allow planners and managers to consider how LULC change in one area in the watershed can cause sedimentation problems at other locations.
17.4 The Model
The Sediment Retention model provides the user with a tool for calculating the average annual soil loss from each parcel of land, determining how much of that soil may arrive at a particular point of interest, estimating the ability of each parcel to retain sediment, and assessing the cost of removing the accumulated sediment on an annual basis. An important determinant of soil retention capacity is land use and land cover. To identify a land parcel’s potential soil loss and sediment transport, the InVEST Avoided Reservoir Sedimentation model uses the Universal Soil Loss Equation (USLE) (Wischmeier & Smith 1978) at the pixel scale, which integrates information on LULC patterns and soil properties, as well as a digital elevation model, rainfall and climate data. The pixel-scale calculations allow us to represent the heterogeneity of key driving factors in water yield such as soil type, precipitation, vegetation type, etc. However, the theory we are using as the foundation of this set of models was developed at the sub-watershed to watershed scale. We are only confident in the interpretation of these models at the sub-watershed scale, so all outputs
17.3. Introduction 284 InVEST User Guide, Release 2.5.6
are summed and/or averaged to the sub-basin scale. We do continue to provide pixel-scale representations of some outputs for calibration and model-checking purposes only. These pixel-scale maps are not to be interpreted for understanding of hydrological processes or to inform decision making of any kind. This model can also be used to value the landscape vis-a-vis maintaining water quality or avoiding reservoir sedi- mentation. In the water quality maintenance case, the model uses additional information on water quality standards and treatment costs to value the ability of each sub-watershed to reduce treatment costs. In the reservoir mainte- nance case, the model uses additional data on reservoir location and the avoided cost of sediment removal to value a sub-watershed’s capacity to keep sediment out of reservoirs.
17.4.1 How it works
First, we estimate the potential for soil loss based on geomorphological and climate conditions. The model is based on the USLE, and represents the first four factors in the equation (rainfall erosivity, soil erodibility, and the length-slope factor). This part of the model accounts for two key relationships. In areas where rainfall intensity is high, there is a high chance that soil particles will become detached and transported by overland runoff. Also, in areas where the soil has a high proportion of sand, the erodibility is high which means soil particles are easily detached from the soil pack and transported by overland runoff. The Universal Soil Loss Equation (USLE) provides the foundation of the biophysical step of the InVEST sediment retention model. USLE = R × K × LS × C × P (from Wischmeier & Smith, 1978) where R is the rainfall erosivity, K is the soil erodibility factor, LS is the slope length-gradient factor, C is the crop- management factor and P is the support practice factor. The Slope Length Factor (LS) is one of the most critical parameters in the USLE. Slope length is the distance from the origin of overland flow along its flow path to the location of either concentrated flow or deposition. It reflects the indirect relationship between slope and land management (terracing, ditches, buffers, barriers). The LS factor is essentially the distance that a drop of rain/sediment runs until its energy dissipates. It represents a ratio of soil loss under given conditions compared to a reference site with the “standard” slope of 9% and slope length of 72.6 feet. The steeper and longer the slope is, relative to the conditions of the reference site, the higher the risk for erosion will be (for more information see http://www.omafra.gov.on.ca/english/engineer/facts/00-001.htm). The estimates of slope-length are based on methodology in a model called N-SPECT such that abrupt changes in slope result in length cutoffs. Adjustments are necessary when slope is greater than 9% and slope length is different than 72.6 feet (22.12m). In the model, different LS equations are automatically used for slope conditions that differ from the standard reference site conditions of the USLE equation development. The slope threshold that the model uses to switch between the follwoing two equations is specified as a model input and depends on the local geomorphology and watershed characteristics.. For low slopes: ! flowacc · cellsizenn sin(slope · 0.01745)1.4 LS = ∗ 1.6 22.13 0.09
0.5, slope ≥ 5% 0.4, 3.5 < slope < 5% nn = 0.3, 1 < slope ≤ 3.5% 0.2, slope ≤ 1% where flowacc is accumulated water flow to each cell and cellsize is the pixel size or the grid resolution (10m, 30m, 90m, etc.).
17.4. The Model 285 InVEST User Guide, Release 2.5.6
For high slopes: We use the following equation, defined by Huang and Lu (1993) for areas with slopes higher than the threshold identified by the user:
LS = 0.08λ0.35prct_slope0.6
cellsize, flowdir = 1, 4, 16, or 64 λ = 1.4 · cellsize, otherflowdir where prct_slope is the pixel’s percent slope and flowdir is the flow direction of the pixel
Calculation of Potential Soil Loss
We estimate the ability of vegetation to keep soil in place on a given pixel by comparing erosion rates on that pixel to what erosion rates would be on that pixel with no vegetation present (bare soil). The bare soil estimate is calculated as follows:
RKLS = R × K × LS
Erosion from the pixel with existing vegetation is calculated by the USLE equation:
USLE = R × K × LS × C × P
Avoided erosion (sediment retention) on the pixel is then calculated by subtracting USLE from RKLS. Vegetation does not only keep sediment from eroding where it grows. It also traps sediment that has eroded upstream. The USLE equation overlooks this component of sediment dynamics, so we attempt to account for it as follows. All soil that the USLE equation estimates will erode is routed downstream via a flowpath. We estimate how much of the sediment eroded on all pixels will be trapped by downstream vegetation based on the ability of vegetation in each pixel to capture and retain sediment. The model also determines the total sediment load exported that reaches the stream from each pixel on the landscape. The table below describes how the removal of sediment by vegetation along hydrologic flowpaths is done:
17.4. The Model 286 InVEST User Guide, Release 2.5.6
The total retained sediment (sretx) is equal to the sum of the sediment removed by the pixel itself and the sediment removed through routing filtration. The model provides the option to consider two services associated with the retention of sediments on the landscape; improved water quality and avoided sedimentation of reservoirs. When considering improved water quality, there may be an allowed annual amount of sediment load for the water body of interest (just as in the water purification model for nutrients). This annual load may be specified by national or local drinking water standards. We subtract this annual allowed load in the service step because people would not receive benefit from retention of sediment upstream of systems that have annual loads below this threshold. We assume that each pixel on the landscape gets an equal proportion of this allowance in the following calculation:
wq_annload sed_ret_wq = sret − x x contrib
where sretx is the total retained sediment calculated above, wq_annload is the annual allowed sediment load and contrib is the number of pixels in the watershed. When considering avoided sedimentation of reservoirs, there is usually an engineered reservoir dead volume, or space built in to the reservoir to capture sediment and avoid the loss of reservoir capacity over time. Because this space is specifically constructed to catch sediment and avoid costs associated with dredging, people do not receive benefit from the landscape’s ability to slow erosion until this dead volume is filled. To account for this and avoid over-valuing this service, we subtract any engineered dead volume in the service step. This calculation is made as follows:
dr_deadvol × 1.26 sed_ret_dr = sret − x x dr_time × contrib
where dr_deadvol is the engineered dead volume of the reservoir, 1.26 is a constant representing the density of sedi- ment in tons m-3, dr_time is the remaining lifetime of the reservoir and contrib is the number of pixels in the watershed. The model then sums (sret_sm_dr; sret_sm_wq) and averages (sret_mn_dr; sret_mn_wq) the sediment export and retention per pixel to the sub-watersheds and provides separate outputs for water quality and dredging. The valuation model uses the cost of sediment removal entered by the user to determine the avoided cost of dredging and/or water quality treatment. . The following equation is used to determine the value each sub-watershed contributes to reservoir maintenance by helping to avoid erosion.
T −1 X 1 sed_V alue = Cost(s) × sret_sm × s (1 + r)t t=0
sed_V alues is the present value of sediment retention on sub-watershed s over T years, where T indicates the period of time over which the LULC pattern is constant (for water quality valuation) or the length of the reservoir life (for dredging valuation), sret_sm is is the total sediment retention adjusted for for either dredging (sret_sm_dr) or water quality (sret_sm_wq), Cost(s) is the marginal cost of sediment removal for either the service of dredging or water quality treatment and r is the discount rate. The Cost(s) may vary across reservoirs or water treatment facilities if different technologies are employed for sediment removal. If this is the case, the user may input reservoir- or plant- specific removal costs. The marginal cost of sediment removal should be measured in units of monetary currency per cubic meter (i.e. $ m-3).
17.4. The Model 287 InVEST User Guide, Release 2.5.6
17.4.2 Limitations and simplifications
Although the USLE method is a standard way to calculate soil loss, it has several limitations. The USLE method predicts erosion from sheet wash alone (erosion from plains in gentle slopes) (FAO 2002). Rill-inter-rill, gullies and/or stream-bank erosion/deposition processes are not included in this model. As such, it is more applicable to flatter areas because it has only been verified in areas with slopes of 1 to 20 percent. Moreover, the relationship between rainfall intensity and kinetic energy may not hold in mountainous areas because it has only been tested in the American Great Plains. Finally, the equation considers only the individual effect of each variable. In reality, some factors interact with each other, altering erosion rates. Another simplification of the model is the grouping of LULC classes because the model’s results are highly sensitive to the categorization of LULC classes. If there is a difference in land use between two areas within the same broad LULC category, it is recommended to create two LULC categories. For example, if all forest is combined into one LULC class, the difference in soil retention between an old growth forest and a newly planted forest is neglected. More generally, where there is variation across the landscape that affects a USLE parameter, the LULC classes should reflect that variation. Third, the model relies on retention or filtration efficiency values for each LULC type. However, there are often few data available locally for filtration rates associated with local LULC types. Data from other regions may be applied in these cases, but may misrepresent filtration by local LULC types. Additionally, the model may not accurately depict the sedimentation process in the watershed of interest since the model is based on parameterization of several different equations and each parameter describes a stochastic process. Due to the uncertainty inherent in the processes being modeled, it is not recommended to make large-scale area decisions based on a single run of the model. Rather, the model functions best as an indicator of how land use changes may affect the cost of sediment removal, and like any model is only as accurate as the available input data. A more extensive study may be required for managers to calculate a detailed cost-benefit analysis for each reservoir site. Another assumption is that sediment retention upstream from a reservoir is valuable only if sediment delivery impacts reservoir function, which incurs a cost. If sediment is not removed from a reservoir, the model does not assign a value to the sediment retention service. In this case, the user may assign a value to upstream sediment retention based on an assumed trajectory of sediment deposition at the reservoir. This method is explained below and it not included in this model. As noted above, we are only modeling sheetwash erosion, meaning that our estimate of annual reservoir sedimentation will be less than actual sedimentation rates. Nonetheless, it is possible to use information about the sediment volume in the reservoir at time t, Vt, and the volume at which reservoir function will be impacted, VD, to estimate the time period over which sediment removal will occur. If the user is able to provide accurate estimates of Vt and VD, then it is likely that information about annual deposition is available as well. Let SEDDEPt represent the total volume of sediment (USLE) assumed to reach the reservoir in a given year. Then we can model the time 0 path of sediment as Vt+1 = SEDDEPt + Vt, and we can define the year at which removal will commence, t , as the first period for which Vt > VD. In this case, let the value of sediment retention on the upstream parcel x be given by PT −1 SEDREMjx×MCd PVSRx∈d = t=t0 (1+r)t where, PVSRx is the present value of sediment retention on pixel x over T years, where T indicates the period of time over which the LULC pattern is constant or the length of the reservoir life length. SEDREMx is the sediment removed by the LULC on pixel x. MC is the marginal cost of sediment removal. r is the discount rate. The accuracy of the sediment retention value is limited by two factors. First, it is limited by the quality of information of the cost of sediment removal. Up-to-date estimates of sediment removal costs for an area may be difficult to find. The user may be limited to using an outdated average value from other locations and for a different type of reservoir. Second, the accuracy of the model is limited by the user’s ability to calibrate it with actual sedimentation data. The model allows for a calibration constant to be applied and adjusted via the Sediment Delivered output. This can greatly improve the model, but only if the user has access to reliable sedimentation data for the watershed(s) of interest.
17.4. The Model 288 InVEST User Guide, Release 2.5.6
17.5 Data needs
Here we outline the specific data used by the model. See the Appendix for detailed information on data sources and pre-processing. For all raster inputs, the projection used should be defined, and the projection’s linear units should be in meters. 1. Digital elevation model (DEM) (required). A GIS raster dataset with an elevation value for each cell. Make sure the DEM is corrected by filling in sinks, and if necessary ‘burning’ hydrographic features into the elevation model (recommended when you see unusual streams.) To ensure proper flow routing, the DEM should extend beyond the watersheds of interest, rather than being clipped to the watershed edge. See the Working with the DEM section of this manual for more information. Name: File can be named anything, but no spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with elevation value for each cell given in meters above sea level. Sample data set: \InVEST\Base_Data\Freshwater\dem 2. Rainfall erosivity index (R) (required). R is a GIS raster dataset, with an erosivity index value for each cell. This variable depends on the intensity and duration of rainfall in the area of interest. The greater the intensity and duration of the rain storm, the higher the erosion potential. The erosivity index is widely used, but in case of its absence, there are methods and equations to help generate a grid using climatic data. See the Appendix for further details. Name: File can be named anything, but no spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with a rainfall erosivity index value for each cell given in MJ*mm*(ha*h*yr)-1. Sample data set: \InVEST\Base_Data\Freshwater\erosivity 3. Soil erodibility (K) (required). K is a GIS raster dataset, with a soil erodibility value for each cell. Soil erodibility, K, is a measure of the susceptibility of soil particles to detachment and transport by rainfall and runoff. Name: File can be named anything, but no spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with a soil erodibility value for each cell. K is in T.ha.h. (ha.MJ.mm)-1. Sample data set: \InVEST\Base_Data\Freshwater\erodibility 4. Land use/land cover (LULC) (required). LULC is a GIS raster dataset, with an integer LULC code for each cell. Name: File can be named anything, but no spaces in the name and less than 13 characters. Format: Standard GIS raster file (e.g., ESRI GRID or IMG), with an LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.) These codes must match LULC codes in the Biophysical table (see below). Sample data set: \InVEST\Base_Data\Freshwater\landuse_90 5. Watersheds (required). A shapefile of polygons. This is a layer of watersheds such that each watershed contributes to a point of interest where water quality will be analyzed. See the Working with the DEM section for information on creating watersheds. Name: File can be named anything, but avoid spaces. Format: Standard GIS shapefile , with unique integer values for each watershed in the ws_id field Sample data set: \InVEST\Base_Data\Freshwater\watersheds.shp
17.5. Data needs 289 InVEST User Guide, Release 2.5.6
6. This option has been removed for the 3.0 version of the sediment model Sub-watersheds (required).A shapefile of polygons. This is a layer of sub-watersheds, contained within the Watersheds (described above) which contribute to the points of interest where water quality will be analyzed. See the Working with the DEM section for information on creating sub-watersheds. Due to limitations in ArcMap geoprocessing, the maximum size of a sub-watershed that can be used in the Sediment Retention model is approximately the equivalent of 4000x4000 cells, with cell size equal to the smallest cell size of your input layers. 7. Biophysical table (required). A table containing model information corresponding to each of the land use classes. NOTE: these data are attributes of each LULC class, not each cell in the raster map. Name: Table names should only have letters, numbers and underscores, no spaces. File type: *.dbf or *.mdb Rows: Each row is a land use/land cover class. Columns: Each column contains a different attribute of each land use/land cover class and must be named as follows: 1. lucode (Land use code): Unique integer for each LULC class (e.g., 1 for forest, 3 for grassland, etc.), must match the LULC raster above. 2. LULC_desc: Descriptive name of land use/land cover class (optional) 3. usle_c: Cover-management factor for the USLE. Note, the ArcGIS version requires the final P and C values given in the table should each be multiplied by 1000. The InVEST 3.0 version requires that P and C are stored in their original floating values. For example, if P=0.2, the ArcGIS version requires the value to be stored as 200 in the table; the 3.0 version requires 0.2. 4. usle_p: Support practice factor for the USLE. Note, the ArcGIS version requires the final P and C values given in the table should each be multiplied by 1000. The InVEST 3.0 version requires that P and C are stored in their original floating values. For example, if P=0.2, the ArcGIS version requires the value to be stored as 200 in the table; the 3.0 version requires 0.2. 5. sedret_eff : The sediment retention value for each LULC class, as an integer percent between zero and 100. This field identifies the capacity of vegetation to retain sediment, as a percentage of the amount of sediment flowing into a cell from upslope. In the simplest case, when data for each LULC type are not available, a value of 100 may be assigned to all natural vegetation types (such as forests, natural pastures, wetlands, or prairie), indicating that 100% of sediment is retained. An intermediary value also may be assigned to features such as contour buffers. All LULC classes that have no filtering capacity, such as pavement, can be assigned a value of zero. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Biophysical_Models 8. Threshold flow accumulation (required). The number of upstream cells that must flow into a cell before it’s considered part of a stream. Used to define streams from the DEM. The model’s default value is 1000. If the user has a map of streams in the watershed of interest, he/she should compare it the Output\Pixel\v_stream map (output of the model). This value also needs to be well estimated in watersheds where ditches are present. This threshold expresses where hydraulic routing is discontinued and where retention stops and the remaining pollutant will be exported to the stream. 9. Slope threshold (required). An integer slope value describing landscape characteristics such as slope manage- ment practices including terracing and slope stabilization techniques. It depends on the DEM resolution and the terracing practices used in the region. In many places, farmers cultivate slopes without any terracing or slope stabilization up to a certain slope, then start implementing these practices or cease agriculture. This slope, where practices stop or switch to terracing or stabilization, should be entered as the slope threshold. The threshold was introduced, along with the alternative LS equation, after application of our model in China in a very steep region of the Upper Yangtze River basin. There, the model performed well when we used a slope threshold of 75% which indicates that agriculture extended into very steep sloping areas, which was the case. In an application in the Cauca Valley, Colombia (in the high Andes), we have used a slope threshold of 90%, basically turning
17.5. Data needs 290 InVEST User Guide, Release 2.5.6
off the alternative slope equation, and the model has performed well there with this approach. If you are unsure of the value to use here, we recommend running the model at least twice, once with the default 75% value and once with 90% and comparing results. If the results are very different (e.g. the model is very sensitive to this input in your region) we recommend finding at least one observation to compare outputs to to guide the decision on the value to use here. 10. Sediment valuation table (required for valuation). A table containing valuation information for each of the reservoirs. There must be one row for each watershed in the Watersheds layer. Name: Table names should only have letters, numbers and underscores, no spaces. File type: *.dbf or *.mdb for ArcGIS models, the standalone model requires a .csv file Rows: Each row is a reservoir or structure that corresponds to the watersheds shapefile. Columns: Each column contains a different attribute of each reservoir and must be named as follows: 1. ws_id (watershed ID): Unique integer value for each reservoir, which must correspond to values in the Watersheds layer. 2. dr_cost: Cost of sediment dredging in $ (Currency) / m3 removed. Floating point value. Used for valuing sediment retention for dredging. 3. dr_time: Integer time period to be used in calculating Present Value (PV) of removal costs. This time period should be the remaining designed lifetime of the structure. For instance, if you are using an LULC map for the year 2000 and a reservoir of interest was designed in 1950 for a 100-year lifetime, the time period entered here should be 50 years. Used for valuing sediment retention for dredging. 4. dr_disc: The rate of discount over the time span, used in net present value calculations. Used for valuing sediment retention for dredging. Floating point value. 5. wq_cost: Cost of removing sediment for water quality in $ (Currency) / m3 removed. Floating point value. Used for valuing sediment retention for water quality. 6. wq_time: Integer time period to be used in calculating Present Value (PV) of removal costs. This time period should be the remaining designed lifetime of the structure. For instance, if you are using an LULC map for the year 2000 and a reservoir of interest was designed in 1950 for a 100-year lifetime, the time period entered here should be 50 years. Used for valuing sediment retention for water quality. 7. wq_disc: The rate of discount over the time span, used in net present value calculations. Used for valuing sediment retention for water quality. Floating point value. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Sediment_Valuation 11. Sediment threshold table (required) A table containing annual sediment load threshold information for each of the reservoirs. There must be one row for each watershed in the Watersheds layer. Name: Table names should only have letters, numbers and underscores, no spaces. File type: *.dbf or *.mdb for ArcGIS models, the standalone model requires a .csv file Rows: Each row is a reservoir or structure that corresponds to the watersheds layer. Columns: Each column contains a different attribute of each reservoir and must be named as follows: 1. ws_id (watershed ID): Unique integer value for each reservoir, which must correspond to values in the Watersheds layer. 2. dr_time: Integer time period corresponding to the remaining designed lifetime of the reservoir (if assessing avoided sedimentation) or the expected time period over which the land use will remain relatively constant. For reservoir sedimentation, if you are using an LULC map for the year 2000
17.5. Data needs 291 InVEST User Guide, Release 2.5.6
and a reservoir of interest was designed in 1950 for a 100-year lifetime, the time period entered here should be 50 years. 3. dr_deadvol: The volume of water below the turbine. It is a design dimension below which water is not available for any use and it’s designed to store (deposit) sediment without hindering turbine and reservoir hydropower functions. Used for calculating service in biophysical terms and valuing retention for dredging. Given in cubic meters. 4. wq_annload: Allowed annual sediment loading, used for valuing sediment retention for water qual- ity. This could be set by national or local water quality standards. Given in metric tons. Sample data set: \InVEST\Base_Data\Freshwater\Water_Tables.mdb\Sediment_Threshold
17.6 Running the Model
The Avoided Reservoir Sedimentation model maps the soil loss, sediment exported, sediment retained, and value of sediment retention on the landscape. This model is structured as a toolkit which has two tools. The first tool, Soil Loss, produces multiple outputs, including USLE, sediment retained by the landscape and sediment exported to the stream. Some of these output values feed into the next portion of the model, the Valuation tool, which calculates sediment retention value. By running the tool, three folders will automatically be created in your workspace (you will have the opportunity to define this file path): “Intermediate”, where temporary files are written and which is deleted after each tool run; “Service”, where results that show environmental services are saved (such as sediment retention); and “Output”, where non-service biophysical results are saved (such as sediment export.) Before running the Avoided Reservoir Sedimentation Model, make sure that the InVEST toolbox has been added to your ArcMap document, as described in the Getting Started chapter of this manual. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. • Identify workspace If you are using your own data, you need to first create a workspace, or folder for the analysis data, on your computer hard drive. The entire pathname to the workspace should not have any spaces. All your output files will be saved here. For simplicity, you may wish to call the folder for your workspace ‘Sediment’ and create a folder in your workspace called “Input” and place all your input files here. It’s not necessary to place input files in the workspace, but advisable so you can easily see the data you use to run your model. Or, if this is your first time using the tool and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you installed the InVEST files on your C drive (as described in the Getting Started chapter), you should see a folder called /Invest/Sedimentation. This folder will be your workspace. The input files are in a folder called /Invest/Base_Data/Freshwater. • Open an ArcMap document to run your model. • Find the InVEST toolbox in ArcToolbox. ArcToolbox is normally open in ArcMap, but if it is not, click on the ArcToolbox symbol. See the Getting Started chapter if you don’t see the InVEST toolbox and need instructions on how to add it. • You can run this analysis without adding data to your map view, but usually it is recommended to view your data first and get to know them. Add the data for this analysis to your map using the ADD DATA button and look at each file to make sure it is formatted correctly. Save your ArcMap file as needed. • Click once on the plus sign on the left side of the InVEST toolbox to see the list of tools expand. Next, click on the plus sign next to the InVEST_Sediment toolset. Within the toolset are two tools, Soil Loss and Valuation. You will need to run Soil Loss first to generate layers that will feed into Valuation. • Double click on Soil Loss. An interface will pop up like the one below. The tool shows default file names, but you can use the file buttons to browse instead to your own data. When you place your cursor in each space,
17.6. Running the Model 292 InVEST User Guide, Release 2.5.6
you can read a description of the data requirements in the right side of the interface. Click Show Help if the description isn’t showing by default. In addition, refer to the Data Needs section above for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires you to enter some data. • After you’ve entered all values as required, click on OK. The script will run, and its progress will be indicated by a “Progress dialogue”. • Upon successful completion of the model, you will see new folders in your workspace called “Intermediate”, “Service” and “Output”. These folders contain several raster grids. These grids are described in the next section. • Load the output grids into ArcMap using the ADD DATA button. • You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting PROPERTIES, and then SYMBOLOGY. There are many options here to change the way the file appears in the map. • You can also view the attribute data of many output files by right clicking on a layer and selecting OPEN ATTRIBUTE TABLE. • Now, run the Valuation Tool. Several outputs from the Soil Loss model are inputs to this model, depend- ing on whether dredging, water quality or both are valued: sret_sm_wq (sediment retention for water quality, summed by sub-watershed), sret_sm_dr (sediment retention for dredging, summed by sub-watershed), sedi- ment_watershed.dbf (table of sediment export/retention per watershed) and sediment_subwatershed.dbf (table of sediment export/retention per sub-watershed.) . The interface is below: • When the script completes running, the outputs will be placed into the “Service” folder. A description of the files is below. • Since this model is open source, the user can edit the scripts to modify, update, and/or change equations by right clicking on the script’s name and selecting “Edit...” The script will then open in a text editor. After making changes, click File/Save to save your new script.
17.7 Interpreting Results
The following is a short description of each of the outputs from the Sediment Retention model. Final results are found in the Output and Service folders within the Workspace specified for this model. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix. • Output\usle_mn (tons/ha): Mean potential soil loss per sub-watershed. • Output\usle_sm (tons/sub-watershed, not /ha): Total potential soil loss per sub-watershed. • Output\sediment_watershed.dbf: Table containing biophysical values for each watershed, with fields as fol- lows: – sed_export (tons/watershed, not /ha): Total amount of sediment exported to the stream per watershed. This should be compared to any observed sediment loading at the outlet of the watershed. Knowledge of the hydrologic regime in the watershed and the contribution of the sheetwash yield into total sediment yield help adjust and calibrate this model. – sed_ret_dr/sed_ret_wq (tons/watershed, not /ha): Total amount of sediment retained by the landscape in each watershed. • Output\sediment_subwatershed.dbf: Table containing biophysical values for each sub-watershed, with fields as follows:
17.7. Interpreting Results 293 InVEST User Guide, Release 2.5.6
17.7. Interpreting Results 294 InVEST User Guide, Release 2.5.6
– sed_export (tons/sub-watershed, not /ha): Total amount of sediment exported to the stream per sub- watershed. – sed_ret_dr/sed_ret_wq (tons/sub-watershed, not /ha): Total amount of sediment retained by the landscape in each sub-watershed for either dredging (_dr) or water quality (_wq). • Output\upret_mn (tons/ha): Raster containing the mean amount of sediment retained from sediment origi- nating upstream of each pixel, averaged across pixels in each sub-watershed. Does not include the sediment originating from the pixel itself. • Output\upret_sm (tons/sub-watershed, not /ha): Raster containing the total amount of sediment retained from sediment originating upstream of each pixel, summed across pixels in each sub-watershed. Does not include the sediment originating from the pixel itself. • Service\sret_mn_wq (Sediment Retained) (tons/ha): Raster containing the mean sediment retained on each sub-watershed, including sediment retained that originates upstream as well as sediment that originates on the cell itself. It is adjusted by the water quality sediment allowable threshold. THIS IS THE SUB-WATERSHED MEASURE OF THIS ENVIRONMENTAL SERVICE IN BIOPHYSICAL TERMS. • Service\sret_sm_wq (Sediment Retained) (tons/sub-watershed, not /ha): Raster containing the total sediment retained within each sub-watershed, including sediment retained that originates upstream as well as sediment that originates on the cell itself. It is adjusted by the water quality sediment allowable threshold. THIS IS THE SUB-WATERSHED MEASURE OF THIS ENVIRONMENTAL SERVICE IN BIOPHYSICAL TERMS. • Service\sret_mn_dr (Sediment Retained) (tons/ha): Raster containing the mean sediment retained per cell on each sub-watershed, including sediment retained that originates upstream as well as sediment that originates on the cell itself. It is adjusted by the reservoir dead volume allowance. THIS IS THE SUB-WATERSHED MEASURE OF THIS ENVIRONMENTAL SERVICE IN BIOPHYSICAL TERMS. • Service\sret_sm_dr (Sediment Retained) (tons/sub-watershed, not /ha): Raster containing the total sediment retained within each sub-watershed, including sediment retained that originates upstream as well as sediment that originates on the cell itself. It is adjusted by the reservoir dead volume allowance. THIS IS THE SUB- WATERSHED MEASURE OF THIS ENVIRONMENTAL SERVICE IN BIOPHYSICAL TERMS. • Output\sexp_mn (tons/ha): Raster containing the mean sediment export for each sub-watershed. • Output\sexp_sm (tons/sub-watershed, not /ha): Raster containing the total sediment export within each sub- watershed. • Service\sed_val_dr (Value of Sediment Removal for dredging) (currency/timespan): Raster showing the value per sub-watershed of the landscape for retaining sediment by keeping it from entering the reservoir, thus avoid- ing dredging costs, over the specified timespan. THIS IS THE SUB-WATERSHED MEASURE OF THIS ENVIRONMENTAL SERVICE IN ECONOMIC TERMS. • Service\sed_val_wq (Value of Sediment Removal for water quality) (currency/timespan): Raster showing the value per sub-watershed of the landscape for retaining sediment by keeping it from entering the reservoir, thus avoiding water quality treatment costs, over the specified timespan. THIS IS THE SUB-WATERSHED MEA- SURE OF THIS ENVIRONMENTAL SERVICE IN ECONOMIC TERMS. • Service\sediment_value_watershed.dbf: Table containing economic values for each watershed, with fields as follows: – sed_export/sed_ret_dr/sed_ret_wq: Same as for sediment_watershed.dbf. – sed_val_dr/sed_val_wq (currency/timespan): Value of the watershed landscape for retaining sediment for either dredging (_dr) or water quality (_wq), over the specified timespan. • Service\sediment_value_subwatershed.dbf: Table containing economic values for each sub-watershed, with fields as follows: – sed_export/sed_ret_dr/sed_ret_wq: Same as for sediment_subwatershed.dbf.
17.7. Interpreting Results 295 InVEST User Guide, Release 2.5.6
– sed_val_dr/sed_val_wq (currency/timespan): Value of the sub-watershed landscape for retaining sediment for either dredging (_dr) or water quality (_wq), over the specified timespan. The application of these results depends entirely on the objective of the modeling effort. Users may be interested in all of these results or select one or two. If sediment removal cost information is not available or valuation is not of interest, the user may use a value of one for the cost of sediment removal. This forces a unit cost of sediment removal, which normalizes the cost across the different reservoirs but still allows a relative comparison of scenarios. The following provides more detail on each of the relevant model outputs. The length-slope factor depends solely on the geometry of the landscape, and, as the name infers, is simply a description of the length of the slopes in the watershed. The RKLS is the potential soil loss based on the length-slope factor, rainfall erosivity, and soil erodibility. These are factors that generally cannot be altered by human activity, as they are inherent to the watershed. USLE differs from RKLS in that it takes into account the management practice factor and the cover factor. These are factors that can be altered with land use changes or changes in land management. Examples of changes that can alter the USLE output are forest clear cuts, changing crop type or type of agriculture (no till to tilled), expansion of an urban area, or restoring vegetation along a stream-bank. The model output describes this ‘actual’ soil loss on an annual basis in tons per hectare, summarized in a raster grid over the landscape. The user should understand that this USLE method predicts the sediment from sheet wash alone. Rill-inter-rill, gullies and/or stream-bank erosion/deposition processes are not included in this model. A visit to the watershed and consul- tation of regional research results need to be used to evaluate the portion of sheet wash in the total sediment loading that is used in testing and verifying this model. Total Sediment exported to the outlet of the watershed (sed_export in the output tables) indicates the volume of soil delivered each year. Since this model doesn’t simulate the in-stream processes where erosion and deposition could have a major impact on the sediment exported, the user should pay great attention to their importance while calibrating or adjusting this model. When soil deposition rates are known from observations at interest points, the user can aggregate the sediment export values (tons of sediment) and compare to observations. Remember that USLE only predicts sheet erosion (not landslide or roads induced or channel erosion), so a sediment budget (distribution of observed sediment yield into erosion types) must be performed to compare the correct measured sources of sediment with the model output. The Value of Sediment Removal is a raster grid that displays the present value (in currency per sub-watershed) of sed- iment retention on the landscape. In other words, it is the avoided cost of sediment removal at a downstream reservoir (over the reservoir’s projected lifetime) due to the ability of the landscape to keep sediment in place. This raster grid provides valuable information to the decision maker on the relative importance of each part of the landscape in deter- mining the cost of sediment removal for a particular reservoir. This output allows managers to see which parts of the landscape are providing the greatest value in terms of avoided sediment removal costs. They may want to protect, or at least avoid serious land use change, in these areas. Similarly, when scenarios of future land management are analyzed with this model, the Value of Sediment Removal layer can be used to identify where the benefits of avoided mainte- nance costs will be lost, maintained or improved across the landscape. Summarizing this layer across the landscape can also give an overall sense of the total costs that will be avoided given a particular landscape configuration. The user should keep in mind that the Tier 1 model may not accurately depict the sedimentation process in the user’s watershed of interest. Furthermore, the model is based on parameterization of several different equations, and each parameter describes a stochastic process. Due to the uncertainty inherent in the processes being modeled here, the user should not make large-scale decisions based on a single run of this model. The Sediment Retention model provides a first cut in prioritization and comparison of landscape management alternatives. A more detailed study is required for managers to calculate a specific benefit-cost analysis for each reservoir site. This model functions best as an indicator of how land use changes may affect the cost of sediment removal, and like any model is only as accurate as the available input data.
17.7. Interpreting Results 296 InVEST User Guide, Release 2.5.6
17.8 Appendix: data sources
This is a rough compilation of data sources and suggestions about finding, compiling, and formatting data. This section should be used for ideas and suggestions only. We will continue to update this section as we learn about new data sources and methods. 1. Digital elevation model (DEM) DEM data is available for any area of the world, although at varying resolutions. Free raw global DEM data is available on the internet from the World Wildlife Fund - http://www.worldwildlife.org/freshwater/hydrosheds.cfm. NASA provides free global 30m DEM data at http://asterweb.jpl.nasa.gov/gdem-wist.asp As does the USGS - http://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/Elevation_Products and http://hydrosheds.cr.usgs.gov/. Or, it may be purchased relatively inexpensively at sites such as MapMart (www.mapmart.com). The DEM resolution is a very important parameter depending on the project’s goals. For example, if decision makers need information about impacts of roads on ecosystem services then fine resolution is needed. And the hydrological aspects of the DEM used in the model must be correct. Please see the Working with the DEM section of this manual for more information. 2. Rainfall erosivity index (R) R should be obtained from published values, as calculation is very tedious. For calculation, R equals E (the kinetic energy of rainfall) times I30 (maximum intensity of rain in 30 minutes in cm/hr). Roose (1996) found that for Western Africa R = a * precipitation where a = 0.5 in most cases, 0.6 near the sea, 0.3 to 0.2 in tropical mountain areas, and 0.1 in Mediterranean mountain areas. The following equation is widely used to calculate the R index (http://www.fao.org/docrep/t1765e/t1765e0e.htm):
R = E · I30 = (210 + 89 log10 I30) ∗ I30
E: kinetic energy of rainfall expressed in metric MJ * m/ha/cm of rainfall. I30: maximum intensity of rain in 30 minutes expressed in cm per hour. In the United States, national maps of the erosivity index can be found through the United States Department of Agriculture (USDA) and Environmental Protection Agency (EPA) websites. The USDA published a loss handbook (http://www.epa.gov/npdes/pubs/ruslech2.pdf ) that contains a hard copy map of the erosivity index for each region. Using these maps requires creating a new line feature class in GIS and converting to raster. Please note that conversion of units is also re- quired (multiply by 17.02). We provide a raster version of this map on the InVEST support site http://invest.ecoinformatics.org/shared. The EPA has created a digital map that is available at http://www.epa.gov/esd/land-sci/emap_west_browser/pages/wemap_mm_sl_rusle_r_qt.htm . The map is in a shapefile format that needs to be converted to raster, along with an adjustment in units. 3. Soil erodibility (K) Texture is the principal factor affecting K, but soil profile, organic matter and permeability also contribute. It varies from 70/100 for the most fragile soil and 1/100 for the most stable soil. It is measured on bare reference plots 22.2 m long on 9% slopes, tilled in the direction of the slope and having received no organic matter for three years. Values of 0 – 0.6 are reasonable, while higher values should be given a critical look. K is sometimes found as part of standard soil data maps, or can be calculated from soil properties.
17.8. Appendix: data sources 297 InVEST User Guide, Release 2.5.6
The FAO provides global soil data in their Harmonized World Soil Database: http://www.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML/. Soil data for many parts of the world are also available from the Soil and Terrain Database (SOTER) Programme (http://www.isric.org/projects/soil-and-terrain-database-soter-programme). In the United States free soil data is available from the U.S. Department of Agriculture’s NRCS in the form of two datasets: SSURGO http://soils.usda.gov/survey/geography/ssurgo/ and STATSGO http://soils.usda.gov/survey/geography/statsgo/ . Where available SSURGO data should be used, as it is much more detailed than STATSGO. Where gaps occur in the SSURGO data, STATSGO can be used to fill in the blanks. The soil erodibility should be calculated for the surface soil horizon of each soil component and then a weighted average of the components should be estimated for the soil class. This can be a tricky GIS analysis: In the US soil categories, each soil property polygon can contain a number of soil type com- ponents with unique properties, and each component may have different soil horizon layers, also with unique properties. Processing requires careful weighting across components and horizons. The Soil Data Viewer (http://soildataviewer.nrcs.usda.gov/), a free ArcMap extension from the NRCS, does this soil data processing for the user and should be used whenever possible. The following equation can be used to calculate K (Wischmeier and Smith 1978):
K = 27.66 · m1.14 · 10−8 · (12 − a) + (0.0043 · (b − 2)) + (0.0033 · (c − 3))
In which K = soil erodibility factor (t*ha/MJ*mm) m = (silt (%) + very fine sand (%))(100-clay (%)) a = organic matter (%) b = structure code: (1) very structured or particulate, (2) fairly structured, (3) slightly structured and (4) solid c = profile permeability code: (1) rapid, (2) moderate to rapid, (3) moderate, (4) moderate to slow, (5) slow and (6) very slow. When profile permeability and structure are not available, as is often the case outside the U.S., soil erodi- bility can be estimated based on soil texture and organic matter content with the following table based on Fig. 21 in Roose (1996): 4. Land use/land cover A key component for all water models is a spatially continuous landuse / land cover raster grid. That is, within a watershed, all landuse / land cover categories should be defined. Gaps in data will create errors. Unknown data gaps should be approximated. Global land use data is available from the University of Maryland’s Global Land Cover Facility: http://glcf.umiacs.umd.edu/data/landcover/. This data is avail- able in 1 degree, 8km and 1km resolutions. Data for the U.S. for 1992 and 2001 is provided by the EPA in their National Land Cover Data product: http://www.epa.gov/mrlc/. The simplest categorization of LULCs on the landscape involves delineation by land cover only (e.g., cropland, temperate conifer forest, prairie). Several global and regional land cover classifications are available (e.g., Anderson et al. 1976), and often detailed land cover classification has been done for the landscape of interest. A slightly more sophisticated LULC classification could involve breaking relevant LULC types into more meaningful types. For example, agricultural land classes could be broken up into different crop types or forest could be broken up into specific species. The categorization of land use types depends on the model and how much data is available for each of the land types. The user should only break up a land use type if it will provide more accuracy in modeling. For instance, for the sediment model the user should only break up ‘crops’ into different crop types if they have information on the difference in soil characteristics between crop management values. 5. P and C coefficients The support practice factor, P, accounts for the effects of contour plowing, strip-cropping or terracing relative to straight-row farming up and down the slope. The cover-management factor, C, accounts for
17.8. Appendix: data sources 298 InVEST User Guide, Release 2.5.6
17.8. Appendix: data sources 299 InVEST User Guide, Release 2.5.6
the specified crop and management relative to tilled continuous fallow. Several references on estimating these factors can be found online: • U.S. Department of Agriculture soil erosion handbook http://topsoil.nserl.purdue.edu/usle/AH_537.pdf • USLE Fact Sheet http://www.omafra.gov.on.ca/english/engineer/facts/00-001.htm • U.N. Food and Agriculture Organization http://www.fao.org/docrep/T1765E/t1765e0c.htm Note, the ArcGIS version requires the final P and C values given in the table should each be mul- tiplied by 1000. The InVEST 3.0 version requires that P and C are stored in their original floating values. For example, if P=0.2, the ArcGIS version requires the value to be stored as 200 in the table; the 3.0 version requires 0.2. 6. Vegetation retention efficiencies These values are used to incorporate the effects of natural vegetation that buffer potential water quality impairment downhill from sources. To develop these values, all land class pixels that contain natural vegetation (such as forests, natural pastures, wetlands, or prairie) are assigned high values and vegetation that has no or little filtering value receives a value of zero. All values should fall between 0 and 100. Consult with a hydrologist if not certain about assignment of specific values. 7. Watersheds / Sub-watersheds Watersheds should be delineated by the user, based on the location of reservoirs or other points of interest. Exact locations of specific structures, such as reservoirs, should be obtained from the managing entity or may be obtained on the web at sites such as the National Inventory of Dams (http://nid.usace.army.mil/). Watersheds that contribute to the points of interest must be generated. If known correct watershed maps exist, they should be used. Otherwise, watersheds and sub-watersheds can be generated in ArcMap using a hydrologically-correct digital elevation model. Due to limitations in ArcMap geoprocessing, the max- imum size of a sub-watershed that can be processed by the Nutrient Retention tool is approximately the equivalent of 4000x4000 cells, at the smallest cell size of all input grids. See the Working with the DEM section of this manual for more information on generating watersheds and sub-watersheds. 8. Sediment table The estimated sediment removal cost from the reservoirs will ideally be based on the characteristics of each reservoir and regional cost data. The user should consult managers at the individual reservoirs or a local sediment removal expert. The technology available at each location may vary, and the applicability of the specific technologies depends on the storage capacity/mean annual runoff ratio and the storage capacity/annual sediment yield ratio. Once a range of possible technologies has been established for each reservoir, the model user should investigate past sediment removal projects to determine appropriate costing. This may require calculating to present day value and taking into account that the technology may have improved, reducing the relative cost. If local information is not available, pricing must be estimated using published information. Adjust costs to specific requirements, location, and present day value as needed. 9. Slope Threshold The threshold was introduced, along with the alternative LS equation, after application of our model in China in a very steep region of the Upper Yangtze River basin. There, the model performed well when we used a slope threshold of 75% which indicates that agriculture extended into very steep sloping areas, which was the case. In an application in the Cauca Valley, Colombia (in the high Andes), we have used a slope threshold of 90%, basically turning off the alternative slope equation, and the model has performed well there with this approach. If you are unsure of the value to use here, we recommend running the model at least twice, once with the default 75% value and once with 90% and comparing results. If the results
17.8. Appendix: data sources 300 InVEST User Guide, Release 2.5.6
are very different (e.g. the model is very sensitive to this input in your region) we recommend finding at least one observation to compare outputs to to guide the decision on the value to use here. 10. Dredging and Water Quality annual loading thresholds Gathering information on water quality standards or targets should be part of the formulation of modeling objectives. If the target to be met is a drinking water target, standards may be set by the federal, state or local level (whichever standard is the most stringent). These standards are set for point of use, meaning that the standard at the point of interest, where water supply will be drawn, may be more relaxed than these standards if water treatment is in place. In-situ water quality standards (for rivers, lakes and streams) may also be set at the national, state and local level. They may be the same across all water bodies of the same type (in rivers, for example) or they may vary depending on the established use of the water body or the presence of endangered species. In the U.S. Total Maximum Daily Loads of sediment are typically established by state regulatory agencies in compliance with the Clean Water Act. States report information on TMDLs to the U.S. EPA on specific waterways http://www2.ctic.purdue.edu/kyw/tmdl/statetmdllists.html.
17.9 References
Anderson, J.R., Hardy E., Roach, J., and Witmer, R. 1976. A Land Use and Land Cover Classification System For Use with Remote Sensor Data: Geological Survey Professional Paper 964. Edited by NJDEP, OIRM, BGIA, 1998, 2000, 2001, 2002, 2005. Desmet, P., and G. Govers (1996) A GIS procedure for automatically calculating the USLE LS factor on topographi- cally complex landscape units. Journal of Soil and Water Conservation, vol. 51, no. 5, pp. 427-433. FAO. 2002., FAOSTAT Homepage of Food and Agriculture Organization of the United Nations, Online 2008, 9/11. Huang Yanhe and Lu Chenglong. 1993. Advances in the application of the Universal Soil Loss Equation (USLE) in China. Journal of Fujian Agricultural College (Natural Science Edition) 22 (1): 73 ~ 77. Roose, E. 1996, Land Husbandry -Components and strategy. 70 FAO Soils Bulletin, Food & Agriculture Organization of the UN, Rome, Italy. Wischmeier, W.H. & Smith, D. 1978, Predicting rainfall erosion losses: a guide to conservation planning. USDA-ARS Agriculture Handbook , Washington DC.
17.9. References 301 CHAPTER EIGHTEEN
MANAGED TIMBER PRODUCTION MODEL
18.1 Summary
An important environmental service provided by forests is the production of timber. This model analyzes the amount and vol- ume of legally harvested timber from natural forests and managed plantations based on harvest level and cycle. The valuption model estimates the economic value of timber based on the market price, harvest and management costs and a discount rate. and calculates its economic value. Limitations of the model include assump- tions that timber harvest production, frequency, prices, and costs are constant over time.
18.2 Introduction
Commercial timber production is a valuable commodity provided by forests, with the potential to generate significant revenue for those with legal rights to harvest. The scale and nature of timber production varies from large privately-operated single-species plantations to small community-managed harvests from natural forest that retains its ecological structure and function. Whether timber production occurs on a managed plantation or a natural forest, managing the intensity and rate of timber harvest is critical to sustaining this service, as well as the supply and value of other services provided by forests, such as water purification, carbon sequestration, and bush meat habitat. Maximizing profits requires information about the volume and species of wood removed in each harvest period, timber prices, and management costs.
18.3 The Model
The model is designed for cases where an entity (e.g., a government, a tribe, a community, a private timber company) has a formally recognized right to harvest roundwood from a forest. According to FAOSTAT (http://faostat.fao.org/), roundwood is wood in its natural state as felled, or otherwise harvested, with or without bark, round, split, roughly squared or in other forms. It comprises all wood obtained from removals. This model’s output maps the net present values of forests’ legally recognized harvests over some user-defined time interval. This model is very simple and designed for cases where little data on harvest practices and tree stand management exists. If you have access to detailed harvest and forest management data, you may want to use an alternative model.
302 InVEST User Guide, Release 2.5.6
Timber harvest by entities that do not have a formally recognized harvesting right is not accounted for in this model. This type of wood harvest, whether it is illegal or occurs in forest areas where property rights are either not defined or not well enforced, is dealt with in the Open Access Timber and Non-Timber Products Model (to be released soon).
18.3.1 How it works
This model can be used in one of two ways. First, it can be used to model the expected value of a stream of harvests from a forest plantation over a user-defined time interval. A forest plantation is typically managed in such a way that merchantable or usable wood can be harvested at regular periods over an indefinite period. Three characteristics of a plantation forest are: 1) species mix has been reduced to a single or a few of the fastest growing species; 2) the oldest wood in the plantation is harvested and the rest of the wood is left to mature; 3) the areas of a plantation that have been clear-cut are replanted with the managed species soon after the clear-cut; and 4) a more or less even distribution of tree ages (e.g., if the oldest trees in the stand are 20 years old, a quarter of the stand is 1-5 year old, a quarter of the stand is 6-10 years old, a quarter of the stand is 11-15 years old, and a quarter of the stand is 16-20 years old). Second, the InVEST Managed Timber Production Model can be used to calculate the expected value of timber harvests from primary, natural forests. By primary, natural forests we mean areas that, at least at the beginning of a harvest cycle, retain much of their natural structure and function. These could include forests that, at least at the beginning of a harvest cycle, are being used by local communities and tribes for small-scale timber and non-timber forest product harvest. In some cases these forests may become subject to large scale timber harvest because they are to transition to more managed forests (i.e., forest plantations as described above) or some other non-forest development that requires a clear-cut, such as agricultural or residential expansion. (This does not include forests that are slashed and burned, given that the felled and burned wood is not used to create a product). In other cases, concessions to clear-cut certain areas of a natural forest or selectively log a natural forest may be held by entities. In these cases an altered version of the natural forest would remain on the landscape into the future. Examples of this type of harvest include logging of rainforests in the Amazon or Malaysia for land conversion or in Indonesia to establish palm plantations, and selective clear-cutting of rainforests in Malaysia. The model runs on a vector GIS dataset that maps parcels on the landscape that are, or are expected to be, used for legal timber harvest over a user-defined time period. These timber parcels can include a whole forest or just part of a forest. In any case a parcel should only include the portion of a forest that is formally designated, zoned, or managed for harvest. Each timber harvest parcel is described by its harvest levels (Harv_mass and Perc_Harv in the production table; see Data Needs section below), frequency of harvest (Freq_harv), and harvest and management (or maintenance) costs (Harv_cost and Maint_cost, respectively) (Fig. 1). The timber parcel map can either be associated with a current (sometimes referred to as “base”) L map is given by yr_cur) or with some future LULC scenario map (where the year associated with the future LULC map is given by yr_fut). If the timber parcel map is associated with the current LULC map the model calculates, for each timber parcel, the net present value (NPV) of harvests that occurred between the current year and some user-defined date, assuming that harvest practices and prices are static over the time interval modeled. If the timber parcel map is associated with a future scenario LULC map the model calculates, for each timber parcel, the NPV of harvests that occurred between the future date and some user-defined date, again assuming that harvest practices and prices do not change over the user-defined time interval. The model produces the NPV of harvests in the currency of either the current year or future year, depending on whether the user inputs a current or future LULC map. For example, if the selected year for the future scenario is 2050 and the dollar is the currency used to value timber harvests, then the NPV of harvests from 2050 to some user-defined later than 2050 is given in year 2050 dollars.
Limitations and simplifications
This model assumes that the percent of the forest harvested each harvest period, the mass of timber harvested each harvest period, the frequency of each harvest period, and harvested related prices and costs remain constant in each timber parcel over the user-defined time period. In reality, each of these variables can change from year to year. For example, the mix of species harvested from a forest could change from one harvest period to the next and this could affect everything from the amount of wood harvested to the composite price received for the timber. In addition,
18.3. The Model 303 InVEST User Guide, Release 2.5.6
Figure 18.1: Figure 1. Conceptual diagram of the Managed Timber Production model. Parameters represented in color are included in the model, while those in gray are not.
18.3. The Model 304 InVEST User Guide, Release 2.5.6 un-modeled disturbances, such as forest fires or disease, or occasional managed thinning can have a major impact on harvest levels from a forest parcels. Some of these limitations can be addressed by constraining the length of the time period used to assess harvests in parcels. For example, if the current year is 2000 and only the expected harvests until 2010 are valued, any unaccounted changes in timber harvest management or price changes may be minor. At this point a future 2010 LULC and timber parcel map could be evaluated with the timber model looking 10 years ahead again, from 2010 to 2020. The future timber parcel map could include any changes in timber management and prices that occurred between 2000 and 2010. This process could be repeated for successive decades until, for example, 2050. Successive model runs with decadal time intervals until 2050, and the ability to change harvesting behavior and prices, will better approximate harvesting practices on the landscape than just running the model once from 2000 to 2050. Further, given the expected variation in harvest management practices and prices over the modeled time interval, it is suggested that the user use mean values for each model input. The mean is typically the best summary of the distribution of expected values for a variable. For example, if it is known that harvests from a timber parcel over time will involve various species it is possible to set the timber price for that parcel equal to the average expected price for all harvested species.
18.4 Data needs
The model requires a GIS polygon file (a vector database) demarcating timber parcels. Unique timber parcels can be distinguished by differences in the percent of the parcel harvested each harvest period, the mass of wood removed each harvest period, the species of trees removed, or the costs of managing and harvesting wood from the parcel. These attributes, along with timber prices and the time interval for analysis, can be included as a table in the shapefile or as a separate table. 1. Timber parcels (required). A GIS dataset (vector) that indicates the different timber parcels on the landscape. Each parcel should be given a unique identifier. The dataset should be projected in meters and the projection used should be defined. Name: file can be named anything File type: standard GIS polygon file (e.g., shapefile), with a unique identifier code for each polygon. Rows: each row is a timber parcel. Columns: Each parcel should be identified with a unique ID. Sample data set: \Invest\Timber\Input\plantation.shp 2. Production table (required). A data table of information about the timber parcels on the landscape. Name: file can be named anything File type: *.dbf, or an attribute table as part of the timber parcel map. Rows: each row is a different parcel. Columns: contain an attribute for each parcel and must be named as follows: 1. Parcel_ID: Same as timber parcel ID in #1. IDs must match the parcel IDs used in the polygon map. User must select this field as a model input. 2. Parcl_area: The area of the timber parcel in hectares. 3. Perc_harv: The proportion of the timber parcel area that is harvested each harvest period; units are integer percent. 4. Harv_mass: The mass of wood harvested per hectare (in metric tons (Mg) ha-1) in each harvest period.
18.4. Data needs 305 InVEST User Guide, Release 2.5.6
5. Freq_harv: The frequency of harvest periods, in years, for each parcel. 6. Price: The marketplace value of the wood harvested from the parcel (-1). This price should reflect what is paid to the harvesters at mills or at other timber processing and collection sites. If a harvest includes multiple species, each with its own price, a weighted price should be used, where weights are given by the expected relative mix of the species in the harvest. Any value derived from pre- commercial thins should be included in Maint_cost (see below). 7. Maint_cost: The annualized cost ha-1 of maintaining the timber parcel, if any. Costs may include the periodic costs to replant, treat and thin the stand, plus the cost to harvest, treat slash, and deliver wood to a processing facility. Other costs may include taxes, pest treatments, etc. If commercial thins before the main harvest produce product that has market value, the annual ha-1 value of these harvests should be subtracted from Maint_cost. If the harvest comes from a natural forest that is not managed for timber production Maint_cost may be 0. (Actual stand maintenance costs may vary from year to year in a forest (e.g., in some years portions of a managed stand may have to be thinned prior to harvest and in other years anti-pest measures may have to be employed), an annualized value “smoothes” this temporal variation in maintenance costs.) 8. Harv_cost: The cost (ha-1) incurred when harvesting Harv_mass. 9. T: The number of years from yr_cur or yr_fut that parcel harvests will be valued. If the parcel is in an even age rotation managed plantation, T can be any number, although as we explain below, we recommend against large T. If the harvest is expected to be an immediate one time clear cut T = 1. If a series of clear cuts in a natural forest are occurring or are expected, T can be no greater than the number of years that harvest of the natural stand can continue given Perc_harv and Freq_harv. For example, if a natural stand is going to be replanted as a single species plantation or allowed to regenerate naturally before being harvested again in the future, T for the harvest of the natural stand can be no larger than 7 if Perc_harv = 33.3 and Freq_harv = 3 (assuming a harvest takes place in years 1 (yr_cur or yr_fut depending on the associated LULC map), 4, and 7). 10. Immed_harv: This attribute answers whether a harvest occurs immediately – whether a harvest occurs in yr_cur, or whether the user is evaluating a forest parcel associated with a future LULC scenario occurring in yr_fut. Answer yes (entered as YES or Y) or no (entered as No or N) to whether a harvest should be calculated for yr_cur or yr_fut. If yes, then the NPV of harvest in the parcel includes a harvest in yr_cur, otherwise the first harvest accounted for in the parcel’s NPV occurs Freq_harv years into the into time interval T. 11. BCEF: An expansion factor that translates the mass of harvested wood into volume of harvested wood. The expansion factor is measured in Mg of dry wood per m3 of wood. The expansion factor is a function of stand type and stand age (this factor is know as the biomass expansion factor in the literature). If you do not have data on this expansion factor you can use the BCEFR row in table 4.5 of IPCC (2006). Otherwise, set this expansion factor equal to 1 for each parcel. Sample data set: \Invest\Timber\Input\plant_table.dbf 3. Market Discount Rate (optional – required for valuation). This number is not supplied in a table, but instead is input directly through a tool interface (Labeled “Market discount rate (%)” in the tool interface.) The market discount rate reflects society’s preference for immediate benefits over future benefits (e.g., would you rather receive $10 today or $10 five years from now?). The tool’s default value is 7% per year, which is one of the rates recommended by the U.S. government for evaluation of environmental projects (the other is 3%). However, this rate will differ depending on the country and landscape being evaluated. It can also be set to 0% if so desired. To calculate NPV for a forest parcel a series of equation are used. First, we calculate the net value of a harvest during a harvest period in timber parcel x,
P erc_harv VH = x (P rice × Harv_mass − Harv_cost ) (18.1) x 100 x x x
18.4. Data needs 306 InVEST User Guide, Release 2.5.6
-1 where VHx is the monetary value (ha ) generated during a period of harvest in x, P erc_harvx is the percentage of x that is harvested in each harvest period (converted to a fraction), Pricex is the market price of a Mg of timber extracted -1 from x, Harv_massx is the Mg ha of wood removed from parcel x during a harvest period, and Harv_costx is -1 the cost (ha ) of removing and delivering Harv_massx to a processing facility or transaction point. In general, -1 Harv_massx will be given by the aboveground biomass (Mg ha ) content of the forest stand less any portion of the stand that is left as waste (e.g., stems, small braches, bark, etc.). For example, assume a company plans to clear-cut -1 -1 10% of a native forest block in each harvest period, Pricex is expected to be $10 , Harv_massx is 800 Mg ha , and Harv_costx = $5,000 ha-1. The net value created during a harvest period is given by,
VHx = 0.1 × (10 × 800 − 5000) = 300 (18.2)
A harvest period is a sustained period of harvest followed by a break in extraction. Plantation forests tend to have a harvest period every year. More natural forests may have more intermittent periods of harvest (e.g., a pulse of harvest activity every 3 years). The periodicity of harvest periods in parcel x is given by the variable F req_harvx.
The variable F req_harvx is used to convert the per hectare value of the parcel (math:VH_x) into a stream of net harvest revenues, which is then aggregated and discounted appropriately. Specifically, the NPV (ha-1) of a stream of harvests that engender math:VH_x intermittingly from yr_cur or yr_fut to Tx years after yr_cur or yr_fut is given by:
Tx ru( F req_harv )−1 ! x VH Mait_cost X x Tx−1 x (18.3) NPVx = − sumt=0 r F req_harvx×s r t s=0 1 + 100 1 + 100
where “ru” means any fraction produced by Tx / F req_harvx is rounded up to the next integer, F req_harvx is the frequency (in years) of harvest periods, r is the market discount rate, and Maint_costx is the annualized cost (ha-1) of managing parcel x. Continuing our earlier example, where math:VH_x = 300, if we set F req_harvx = 1 (a harvest period occurs every year), Tx equal to 10 (Tx can be no larger than 10 because the native forest will be completely -1 gone in 10 years given P erc_harvx = 10%), r equal to 7%, and Maint_costx equal to $50 ha , then the NPV of the stream of math:VH_x is,
9 9 X 300 X 50 NPV = − (18.4) x 1.07s 1.07t s=0 t=0
On the other hand, assume F req_harvx = 3 (a 10% harvest of the timber parcel occurs every 3 years) and all other variables are as before, then,
10 ru( 3 )−1 9 X 300 X 50 NPV = − (18.5) x 1.073×s 1.07t s=0 t=0
In other words, a harvest period occurs in years 1 (yr_cur or yr_fut), 4, 7, and 10 with annualized management costs incurred every year (where s = 0 refers to year 1, s = 1 refers to year 4, s = 2 refers to year 7 and s = 3 refers to year 10). Note that when using equation (3) we always assume a harvest period in yr_cur or yr_fut, the next occurs F reqx years later, the next 2 F reqx years later, etc.
Alternatively, if a harvest does not take place in yr_cur or yr_fut, and instead the first one is accounted for F reqx years into the time interval T, then we use the following equation,
rd Tx ( F req_harvx ) Tx−1 ! X VHx X Mait_costx (18.6) NPVx = − t r (F req_harvx×s)−1 1 + r s=1 1 + 100 t=0 100
18.4. Data needs 307 InVEST User Guide, Release 2.5.6
where “rd” means any fraction produced by Tx / F req_harvx is rounded down to the next integer In this case, if F req_harvx = 3 and Tx = 10, then x experiences a harvest period in years 3, 6, and 9 of the time interval.
The selection of Tx and F reqx require some thought. First, if timber parcel x is expected to only experience one immediate harvest period (either in the base year with equation (3) or F reqx-years into the time interval with equation (6)), then set Tx = F reqx = 1. On the other hand, if parcel x is in an even-aged managed rotation, then the value of Tx can be set very high (we assume that harvests can be sustained indefinitely in such systems). However, we recommend against using large Tx values for any x for several reasons. First, in this model, timber price, harvest cost, and management cost are static over time. This may only be a reasonable assumption for short periods of time (e.g., 20 years). Second, in this model, timber management is static over time; again this may only be a reasonable assumption over short periods of time. Third, if natural forests are being transformed into plantations, a large T would require that we begin accounting for the eventual plantation harvests. This complication would make the model less tractable. Note that F reqx Tx for all x.
Finally, the net present value of timber harvest for the entire area of parcel x from the base year to Tx years later is given by TNPVx, where Parcl_areax is the area (ha-1) of parcel x:
TNPVx = P arcl_areax × NPVx (18.7)
The last table entry, BCEFx, is used to transform the total volume of wood removed from a parcel from yr_cur or yr_fut to T years later (TBiomassx). If Immed_harvx = 1 then, P erc_harvx Tx T biomassx = P arcl_areax × × Harv_massx × ru (18.8) 100 F req_harvx
Otherwise, if Immed_harvx = 0 then P erc_harvx Tx T biomassx = P arcl_areax × × Harv_massx × rd (18.9) 100 F req_harvx and 1 T V olumex = T Biomassx × (18.10) BCEFx
Example: Landscape with timber production in five parcels. In this example, the first two timber parcels are managed for timber production on a 45-year even-age rotation (1/45 of the stand is harvested and then replanted each year) in perpetuity, but have different mixes of species and different management costs. Each managed timber parcel is 1000 hectares. The third timber parcel has the same species mix as the second, but 1/4 of the parcel is harvested every 20 years and it will only be managed for at least another 50 years. The fourth polygon is a clear-cut of a 500 ha natural forest that is slated to become a shopping mall. The fifth parcel represents a portion of a mature, primary forest. The parcel in the larger forest that will be used for timber production is 500 ha. It will be systematically clear-cut over the next ten years and then managed as a single species plantation indefinitely (we do not account for the plantation’s expected revenues in this model). Par- Parcl_area Perc_harv Freq_harv Harv_massPrice Maint_costHarv_cost T Immed_harvBCEF cel_ID 1 1000 2.22 1 80 300 190 50 50 Y 1 2 1000 2.22 1 70 200 260 124 50 Y 1 3 1000 25 20 70 200 310 225 50 N 1 4 500 100 1 95 350 180 45 1 Y 1 5 500 20 2 95 400 190 105 10 Y 1
18.4. Data needs 308 InVEST User Guide, Release 2.5.6
18.5 Running the Model
Before running the Timber Model, first make sure that the INVEST toolbox has been added to your ARCMAP doc- ument, as described in the Getting Started chapter of this manual. Second, make sure that you have prepared the required input data files according to the specifications in Data Needs. Specifically, you will need (1) a shapefile or raster file showing the locations of different timber management zones in the landscape; (2) a table with data on harvest frequency and amount, and the price of timber and cost of harvest; and (3) the discount rate for timber, if other than the 7% US government estimate. • Identify workspace If you are using your own data, you need to first create a workspace, or folder for the analysis data, on your computer hard-drive. The entire pathname to the workspace should not have any spaces. All your output files will be dumped here. For simplicity, you may wish to call the folder for your workspace “timber” and create a folder in your workspace called “input” and place all your input files here. It’s not necessary to place input files in the workspace, but advisable so you can easily see the data you use to run your model. Or, if this is your first time using the tool and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you unzipped the InVEST files to your C-drive (as described in the Getting Started chapter), you should see a folder called /Invest/timber. This folder will be your workspace. The input files are in a folder called /Invest/timber/input and in /Invest/base_data. • Open an ARCMAP document to run your model. • Find the INVEST toolbox in ARCTOOLBOX. ARCTOOLBOX is normally open in ARCMAP, but if it is not, click on the ARCTOOLBOX symbol. See the Getting Started chapter if you don’t see the InVEST toolbox and need instructions on how to add it. • You can run this analysis without adding data to your map view, but usually it is recommended to view your data first and get to know them. Add the data for this analysis to your map using the ADD DATA button and look at each file to make sure it is formatted correctly. Save your ARCMAP file as needed.
• Click once on the plus sign on the left side of the INVEST toolbox to see the list of tools expand. Double-click on TIMBER. • An interface will pop up like the one below. The tool indicates default file names, but you can use the file buttons to browse instead to your own data. When you place your cursor in each space, you can read a description of the data requirements in the right side of the interface. In addition, refer to the Data Needs section above for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, it requires you to enter some data. • After you’ve entered all values as required, click on OK. The script will run, and its progress will be indicated by a “Progress dialogue.” • Upon successful completion of the model, you will see new folders in your workspace called “intermediate” and “output.” These folders contain several raster grids. These grids are described in the next section. • Load the output grids into ARCMAP using the ADD DATA button.
• You can change the symbology of a layer by right-clicking on the layer name in the table of contents, selecting PROPERTIES, and then SYMBOLOGY. There are many options here to change the way the file appears in the map. • You can also view the attribute data of output files by right clicking on a layer and selecting OPEN ATTRIBUTE TABLE.
18.5. Running the Model 309 InVEST User Guide, Release 2.5.6
18.5. Running the Model 310 InVEST User Guide, Release 2.5.6
18.5. Running the Model 311 InVEST User Guide, Release 2.5.6
18.6 Interpreting results
18.6.1 Final Results
Final results are found in the Output folder within the Workspace specified for this module. • Timber.shp – The attribute table has three columns. The first column gives each timber parcel’s TNPV. TNPV is the net present economic value of timber production in terms of the user-defined currency. TNPV includes the revenue that will be generated from selling all timber harvested from yr_cur or yr_fut to T years after yr_cur or yr_fut less harvest and management costs incurred during this period. Finally, all monetary values are discounted back to yr_cur or yr_fut‘s present value. Negative values indicate that costs (management and harvest) are greater than income (price times harvest levels). The TBiomass column gives the total biomass (in Mg) of harvested wood removed from each timber parcel from yr_cur or yr_fut to T years after yr_cur or yr_fut (TBiomass from equation (8) or equation (9), depending on the value of Immed_harv). The TVolume column gives the total volume (m3) of harvested wood removed from each timber parcel from yr_cur or yr_fut to T years after yr_cur or yr_fut (TVolume from equation (10)).
18.7 Timber 3.0 Beta
Currently we are working on the next generation platform of InVEST and deploying parts of it as prototype InVEST models. Managed timber has a 3.0 prototype which can be found in the InVEST 3.0 Beta toolbox inside the InVEST 2.5.6 toolbox. Currently it is only supported in ArcGIS 10. New features to the 3.0 version include: • Paramters from previous runs are automatically loaded into the user interface. • Runtime of the model has been improved. Please send feedback or bug reports to [email protected].
18.8 References
Maass, J., P. Balvanera, A. Castillo, GC Daily, HA Mooney, P. Ehrlich, M. Quesada, A. Miranda, VJ Jaramillo, F. Garcia-Oliva, A. Martinez-Yrizar, H. Cotler, J. Lopez Blanco, A. Perez Jimenez, A. Burquez, C. Tinoco, G. Ceballos, L. Barraza, R. Ayala, and J. Sarukhan. 2005. Ecosystem services of tropical dry forests: insights from long-term ecological and social research on the Pacific Coast of Mexico. Ecology and Society 10:17. Nunez, D., L. Nahuelhual, and C. Oyarzun. 2006. Forests and water: The value of native temperate forests in supplying water for human consumption. Ecological Economics 58:606-616. Ricketts, TH. 2004. Tropical Forest Fragments Enhance Pollinator Activity in Nearby Coffee Crops. Conservation Biology 18:1262-1271. Sohngen, B., and S. Brown. 2006. The influence of conversion of forest types on carbon sequestration and other ecosystem services in the South Central United States. Ecological Economics 57:698-708.
18.6. Interpreting results 312 CHAPTER NINETEEN
CROP POLLINATION
19.1 Summary
Seventy-five percent of globally important crops rely either in part or completely on animal pollination. The InVEST pollination model focuses on wild bees as a key animal pollinator. It uses es- timates of the availability of nest sites and floral resources and bee flight ranges to derive an index of bee abundance nesting on each cell on a landscape (i.e., pollinator supply). It then uses flight range information to estimate an index of bee abundance visiting each agricultural cell. If desired, the model then calculates a sim- ple index of the value of these bees to agricultural production, and attributes this value back to source cells. The results can be used to optimize agriculture and conservation investments. Required inputs include a current land use and land cover map, land cover attributes, species of pollinators present, and their flight ranges. The model’s limitations include exclusion of non-farm habitats that may determine pollinator abundance and of the effects of land parcel size. The model also does not account for managed pollinators and pollinator persistence over time.
19.2 Introduction
Crop pollination by bees and other animals is a potentially valu- able ecosystem service in many landscapes of mixed agricultural and natural habitats (Allen-Wardell et al. 1998, Free 1993). Pollination can increase the yield, quality, and stability of fruit and seed crops as diverse as tomato, canola, watermelon, coffee, sunflower, almond, and cacao. Indeed, Klein et al. (2007) found that 87 of 115 globally important crops benefit from animal pollination, a service valued variously in the billions to tens of billions per year globally (Costanza et al. 1997, Losey and Vaughan 2006, Nabhan and Buchmann 1997, Southwick and Southwick 1992). Despite these numbers, it is important to realize that not all crops need animal pollination. Some crop plants are wind (e.g., staple grains such as rice, corn, wheat) or self pollinated (e.g., lentils and other beans), needing no animal pollinators to successfully produce fruits or seeds. Klein et al. (2007) provides a list of crops and their pollination requirements that can help identify whether crops in a region of interest may benefit from wild animal pollinators. Decision-makers can use information on crop pollinators, their abundance across a landscape, and the pollination services they provide to crops in several ways. First, with maps of pollinator abundance and crops that need them, land use planners could predict consequences of different policies on pollination services and income to farmers (for an example, see Priess et al. 2007). Second, farmers could use these maps to locate crops intelligently, given
313 InVEST User Guide, Release 2.5.6 their pollination requirements and predictions of pollinator availability. Third, conservation organizations or land trusts could use the tool to optimize conservation investments that benefit both biodiversity and farmers. Finally, governments or others proposing payment schemes for environmental services could use the results to estimate who should pay whom, and how much.
19.3 The Model
A wide range of animals can be important pollinators (e.g., birds, bats, moths and flies), but bees are the most important group for most crops (Free 1993). As a result, the InVEST Pollination model focuses on the resource needs and flight behaviors of wild bees. Many people think of honeybees, managed in artificial hives, when they think of pollinators, but wild bees also contribute to crop pollination. In fact, for several important crops (e.g., blueberries), native species are more efficient and effective pollinators than honeybees (Cane 1997). These native bees, in addition to feral honeybees living in the wild, can benefit crops without active management of captive hives. This is the pollination service associated with habitat conservation. For bees to persist on a landscape, they need two things: suitable places to nest, and sufficient food (provided by flowers) near their nesting sites. If provided these resources, pollinators are available to fly to nearby crops and pollinate them as they collect nectar and pollen. The model therefore uses information on the availability of nesting sites and flower resources, as well as flight ranges of bees, to map an index of bee abundance across the landscape. In a second step, the model uses this map and bee flight ranges again to predict an index of the number of pollinators likely visiting crops in each agricultural cell on the landscape. If you opt to also estimate value indices, the model then takes a third and fourth step. In the third step, it uses a simplified yield function to translate bee abundance into crop value on each agricultural cell. And in the fourth step, it attributes these cell values back to cells “supplying” these bees. These steps are laid out in more detail below, and the full model description can be found in Lonsdorf et al. (in press).
19.3.1 How it works
The model is based on a land use and land cover (LULC) map, showing both natural and managed land types. This map is divided into a regular grid of square cells, each of which is assigned a single LULC type. For each type, the model requires estimates of both nesting site availability and flower availability (e.g., for bee food: nectar and pollen). These data can be supplied from quantitative field estimates or from expert opinion, and are expressed in the form of relative indices (between 0 and 1). Flower availability can be supplied separately for different seasons if important, and the availability of nesting substrates can be estimated separately for multiple nesting guilds (e.g., ground nesters, cavity nesters). Because bees are proficient flyers, they integrate over several elements of a landscape, moving between nesting habitats and foraging habitats (Ricketts et al. 2006). The distances they typically fly affect both their persistence and the level of service they deliver to farms. The model therefore requires a typical foraging distance for each pollinator species. These data can be supplied from quantitative field estimates (e.g., Roubik and Aluja 1983), proxies such as body size (Greenleaf et al. 2007), or from expert opinion. Using these data, the model first estimates the abundance index of each pollinator species in every cell in the landscape, based on the available nesting sites in that cell and the flowers (i.e., food) in surrounding cells. Flowers in nearby cells are given more weight than distant cells, according to the species’ average foraging range. Since pollinator abundance is limited by both nesting and floral resources, the pollinator abundance for species β index on cell x, Pxβ, is the product of foraging and nesting such that:
−D M mx P αβ m=1 Fjme Pxβ = Nj −D M mx P αβ m=1 e
19.3. The Model 314 InVEST User Guide, Release 2.5.6
where Nj is the suitability of nesting of LULC type j, Fj is the relative amount floral resources produced by LULC type j, Dmx is the Euclidean distance between cells m and x and αβ is the expected foraging distance for the pollinator β (Greenleaf et al. 2007). The result is a map of the abundance index (0-1) for each species, which represents a map of “pollinator supply” (i.e., bees available to pollinate crops). In this sense, this map represents the potential sources of pollination services, but it has not yet incorporated demand. In other words, the landscape may be rich in pollinator abundance, but if there are no bee-pollinated crops on that landscape, those bees will not be providing the service of crop pollination. To make this connection between areas of “supply” and “demand,” the model calculates an abundance index of visiting bees at each agricultural cell, by again using flight ranges of pollinator species to simulate their foraging in nearby cells. Specifically, it sums pollinator supply values in cells surrounding each agricultural cell, again giving more weight to nearby cells. This sum, created separately for each pollinator species at each agricultural site, is an index of the abundance of bees visiting each farm site (i.e., “farm abundance”). We use the foraging framework described in the previous equation to determine the relative abundance of bees that travel from a single source cell x to forage on a crop in agricultural cell o:
−Dox α Pxβe β Poxβ = −D M ox P αβ x=1 e
where Pxβ is the supply of pollinators of species β on cell x, Doxβ is distance between source cell x and agricultural cell o for that species, and αβ is species’ β average foraging distance. The numerator of this equation represents the distance-weighted proportion of the pollinators supplied by cell m that forage within cell o and the numerator is a scalar that normalizes this contribution by the total area within foraging distance (Winfree et al. 2005). The total pollinator abundance on agricultural cell o, Po, is simply the sum over all M cells. This second map represents the relative degree of pollination service at the demand points, or points at which this service is “delivered”: agricultural cells. The actual economic benefit received from pollination depends on how crops grown in each cell respond to pollinators. The model therefore takes two additional (optional) steps to translate farm abundances of pollinators into indices of expected economic value. In lieu of a more detailed agricultural production function, we use a simple saturating crop yield function, which assumes that yield increases as pollinator visitation increases, but with diminishing returns (Greenleaf and Kremen 2006). Crops vary in their dependence on pollinators; some crop species are self-compatible and yield is less dependent on pollination while other species obligately require pollination to generate any yield (Klein et al. 2007). We account for both observations, and thus calculate the expected yield of a crop c on farm o, Yoβ, as:
Poβ Yoβ = 1 − νc + νc Poβ + κc
Where νc represents the proportion of total crop c’s yield attributed only to wild pollination (e.g. νc would be equal to 1 if a crop is an obligately outcrossing species and equal to 0 if the crop species were wind-pollinated). In the denominator of the third term, κc is a half-saturation constant and represents the abundance of pollinators required to reach 50% of pollinator-dependent yield. Once the model has calculated value for each agricultural cell, it redistributes this value back to cells that supplied the relevant pollinators, creating a map of value at the source. First, the model assigns fractions of the cell’s value to each of the bee species, according to their partial contribution to total farm abundance. Then each species’ value is redistributed back to the source cells from which they came using the same distance-weighted relationship described above. Thus source habitats close by provide greater service value than those farther away. Formally, we calculate pollinator service provided to O farms from each m cell, PSm, as:
−D M Y mx P oβm e αβ m=1 Poβm PSxβ = νoPxβ −D M mx P αβ m=1 e
19.3. The Model 315 InVEST User Guide, Release 2.5.6
where Vo represents the crop value in farm cell o. The result is a map of “pollinator service value” that estimates the relative index of economic value of pollinators for agricultural areas. If the simple saturating yield function is deemed too simplistic, one may link this pollination model to InVEST’s agricultural production model that includes other factors such as fertilizer, irrigation, labor, etc. The integration of these two models will give a more appropriate representation of the multiple inputs to agricultural production. It will also be possible to more specifically derive the amount of crop yield provided by wild pollinators (yield contribution) and the net present value of that additional yield. See Lonsdorf et al. (2009) and Lonsdorf et al. (in press) for equations that determine the pollinator supply, farm abundance, and pollinator service value maps.
19.3.2 Limitations and simplifications
The model predicts an abundance index of wild pollinators on agricultural fields (cells) within a landscape, based on the pattern of land cover types and the resources they are estimated to contain for bees. It also converts this abundance into indices of production value and attributes this value to the source cells for pollinators. Like other InVEST models, the Pollination model is extremely simple, but it makes reasonably accurate predictions when compared to field obser- vations of pollinators (Lonsdorf et al. 2009). Nevertheless, with this simplicity come several limitations that must be kept in mind. First, the model predicts only relative patterns of pollinator abundance and pollination value (using indices of 0-1). This is because absolute estimates of nest density, resource availability, and pollinator abundance are rarely available, and yield functions (including pollinator abundance) for many crops are poorly defined. However, relying on relative indices limits our ability to estimate absolute economic values to better inform land-use planning decision-making, often based on cost-benefit analyses. This simplicity is perhaps most limiting in calculating indices of value, both on farms and at the source cells of pollinator supply. With field samples of absolute pollinator abundance, one could calibrate InVEST’s relative indices to predict actual pollinator abundances. And with specific yield functions, one could use these actual abundances to estimate absolute estimates of economic value. This would require, beyond these additional data, custom modeling steps that InVEST does not offer. InVEST does produce, however, the intermediate results necessary to insert these modeling steps. Furthermore, the logic that increasing pollinator abundance and diversity lead to increased yield is supported by previous research (Greenleaf and Kremen 2006). One option for overcoming this limitation is to link this model with an agricultural production model (InVEST or another), which will take pollinator abundance as one input to predict and map agricultural yields. In formal terms, it will use pollination as a factor in a “production function” that relates yields of a given crop to the quantity and quality of various inputs (e.g., water, soil fertility, labor, chemicals, pollination). Using these production functions, it is possible to estimate the proportion of crop productivity that is due to pollination, and thus the economic value of those pollinators. Second, the model does not include the dynamics of bee populations over time, and therefore cannot evaluate whether these populations are sustainable given the current landscape. Instead, the model simply provides a static snapshot of the number of pollinators on each cell in the landscape, given simple estimates of nesting sites and food resources. Some of the factors that influence bee populations, like habitat disturbances and typical population fluctuations, are not captured. Third, the model does not account for the sizes of habitat patches in estimating abundance. For many species, there is a minimum patch size, under which a patch cannot support that species over the long term. There is some evidence that small patches support fewer species of bees (Kremen et al. 2004), but bees can also survive in surprisingly small areas of suitable habitat (Ricketts 2004). Fourth, pollinators are likely to be influenced by fine-scale features in the landscape, which are difficult to capture in typical land-cover data with typical resolutions of 1km or even 30m. For example, small patches of flower resources in an otherwise hostile habitat for bees can provide important food resources, but will not be detected by typical land cover maps. Some bees are also able to nest in small but suitable areas (a single suitable roadside or tree hollow). Using average values of nesting site or flower availability for each land cover type, along with 30m pixels or larger, will therefore not capture these fine scale but important areas of resources.
19.3. The Model 316 InVEST User Guide, Release 2.5.6
Finally, the model does not include managed pollinators, such as honey bees, that are managed in boxed hives and can be moved among fields to pollinate crops. InVEST focuses on the environmental service of pollination by bees living wild in the landscape. Managed pollinators are a technological substitute for this environmental service, much as a water filtration plant is a substitute for purification services by wetlands and other natural systems. Clearly, any natural resource assessment needs to consider the costs and benefits of investments in technology (filtration plants, managed bees) alongside those of investments into natural capital (wetlands, wild bee pollination).
19.4 Data needs
The model uses five forms of input data (three are required, and two are optional): 1. Current land cover map (required). A GIS raster dataset, with a land use and land cover (LULC) code for each cell. The dataset should be projected in meters and the projection should be defined. This coverage must be of fine enough resolution (i.e., sufficiently small cell-size) to capture the movements of bees on a landscape. If bees fly 800 meters on average and cells are 1000 meters across, the model will not fully capture the movement of bees from their nesting sites to neighboring farms. Name: file can be named anything, but avoid spaces (e.g. use lulc_samp_cur) Format: standard GIS raster file (e.g., ESRI GRID or IMG), with a column labeled ‘value’ that designates the LULC class code for each cell (e.g., 1 for forest, 3 for grassland, etc.) The LULC ‘value’ codes must match LULC class codes used in the Land Attributes table described below. The table can have additional fields, but the only field used in this analysis is one for LULC class code. The model also requests three pieces of information about this LULC map, which are optional but will be prompted for in the interface. 1. The year depicted by the LULC map (optional). You can indicate the year of the LULC map, if known, to designate model runs performed at different time periods (i.e., future scenarios). 2. The resolution at which the model should run (optional). You can indicate a coarser resolution than that of the native LULC map to prompt the model to resample at this new resolution and to speed up run time. For example, you could run the model at a 200m resolution with a 30m resolution LULC map. If you leave this line blank, the model will perform the analysis at the same resolution of the native LULC map (i.e., the default). (Note: a resolution that is finer than the native resolution of the raster dataset cannot be defined). 3. Agricultural land cover and land use classes (optional). You can specify LULC classes that represent agricultural parcels dependent upon or that benefit from pollination by bees. Doing so will restrict the calculation of pollinator abundance to only the designated farms. Enter the LULC values in the format 2;9;13;etc. If you do not specify agricultural classes then a farm abundance map will be calculated for the entire landscape (the default). Refer to Klein et al. 2007 for a list of crops and their level of pollinator-dependency. Sample data set: \Invest\Base_Data\Terrestrial\lulc_samp_cur 2. Table of pollinator species or guilds (required). A table containing information on each species or guild of pollinator to be modeled. Guild refers to a group of bee species that show the same nesting behavior, whether preferring to build nests in the ground, in tree cavities, or other habitat features. If multiple species are known to be important pollinators, and if they differ in terms of flight season, nesting requirements, or flight distance, provide data on each separately. If little or no data are available, create a single ‘proto-pollinator,’ with data taken from average values or expert opinion about the whole pollinator community. Name: file can be named anything File Type: *.dbf, Excel worksheets (*.xls, .xlsx), or Ms Access tables (*.mdb, .accdb). If using ArcGIS 9.2x then you will need to use .xls or .mdb files. Excel 2007 (.xlsx) and Ms Access 2007 (.accdb) files will only work with ArcGIS 9.3x.
19.4. Data needs 317 InVEST User Guide, Release 2.5.6
Rows: each row is a unique species or guild of pollinator. Columns: columns contain data on each species or guild. Column order doesn’t matter, but columns must be named as follows (italicized portions of names can be customized for meaning, but must be consistent with names in other tables): 1. Species: Name of species or guild (Note: species names can be numerical codes or names. The model will produce outputs coded by the first 4 characters of each species name (e.g., Andr for Andrena nivalis), thus, each species or guild should be uniquely identifiable at 4 characters. If species or guild are not uniqueluely identifiable at 4 characters then the model will truncate the names at 3 and at a digit). 2. NS_nest1, NS_nest2, etc.: Nesting guilds of each pollinator. Values should be entered either as 0 or 1, with 1 indicating a nesting type that is utilized and 0 indicating a non-utilized nest type. If a pollinator falls within multiple nesting guilds, then indicate 1s for all compatible nest types. Nesting types might be ground nests, tree cavities, etc. 3. FS_season1, FS_season2, etc.: Pollinator activity by floral season (i.e., flight season). Values should be entered on a scale of 0 to 1, with 1 indicating the time of highest activity for the guild or species, and 0 indicating no activity. Intermediate proportions indicate the relative seasonal activity. Activity level by a given species over all seasons should sum to 1. Create a different column for each season. Seasons might be spring, summer, fall; wet, dry, etc. 4. Alpha: average (or typical) distance each species or guild travels to forage on flowers, specified in meters. InVEST uses this estimated distance to define the neighborhood of available flowers around a given cell, and to weight the sums of floral resources and pollinator abundances on farms. You can determine typical foraging distance of a bee species based on a simple allometric relationship with body size (see Greenleaf et al. 2007). Sample data set: \InVEST\Pollination\input\Guild.dbf Example: A hypothetical study with four species. There are two main nesting types, “cavity” and “ground.” Species A is exclusively a cavity nester, species B and D are exclusively ground nesters, and species C uses both nest types. There is only a single flowering season, “Allyear,” in which all species are active. Typical flight distances, specified in meters (Alpha), vary widely among species. Species NS_cavity NS_ground FS_allyear Alpha A 1 0 1 1490 B 0 1 1 38 C 1 1 1 890 D 0 1 1 84 3. Table of land cover attributes (required). A table containing data on each class in the LULC map (as described above in #1). Data needed are relative indices (0-1), not absolute numbers. Data can be summarized from field surveys, or obtained by expert assessment if field data is unavailable. Name: file can be named anything File type: *.dbf, Excel worksheets (*.xls, .xlsx), or Ms Access tables (*.mdb, .accdb). If using ArcGIS 9.2x then you will need to use .xls or .mdb files. Excel 2007 (.xlsx) and Ms Access 2007 (.accdb) files will only work with ArcGIS 9.3x. Rows: each row is a different LULC class. Columns: each column contains a different attribute of each LULC class, and must be named as follows: 1. LULC: Land use and land cover class code. LULC codes match the ‘values’ column in the LULC raster and must be numeric, in consecutive order, and unique. 2. LULCname: Descriptive name of LULC class (optional). 3. N_nest1, N_nest2, etc.: Relative index of the availability of nesting type 1, 2, etc. within each LULC type, on a scale of 0-1 (values do not need to sum to 1 across nesting types). Set the LULC type with
19.4. Data needs 318 InVEST User Guide, Release 2.5.6
the greatest availability of nesting habitat at 1, and give all other land classes a value in proportion to this maximum value. The italicized parts of names must match those in NS_nest1, etc. in the Table of pollinator species or guilds (described in input #2 above). 4. F_season1, F_season2, etc.: Relative abundance (0-1) of flowers in each LULC class for season 1, season 2, etc. There are two aspects to consider when estimate relative floral abundance of each LULC class: % floral abundance or % floral coverage as well as the duration of flowering during each season. For example, a land cover type that comprises 100% of a mass flowering crop that flowers the entire season with an abundance cover of 80% would be given a suitability value of 0.80. A land cover type that flowers only half of the season at 80% floral coverage would be given a floral suitability value of 0.40. Italicized parts of names must match those in FS_nest1, etc. in the Table of pollinator species or guild file (described in input #2 above). Sample data set: \InVEST\Pollination\input\LU.dbf Example: The same hypothetical study with five LULC classes. Class 1 (Forest) contains the maximum availability of sites for both nesting types (“cavity” and “ground”). The five habitat types vary strongly in flower resources in the single (simplified, year-round) flowering season. Note matching column heads between this table and the Table of pollinator species or guilds. LULC LULCname N_cavity N_ground F_allyear 1 Forest 1.0 1.0 1.0 2 Coffee 0.2 0.1 0.5 3 Pasture/grass 0.2 0.1 0.3 4 Shrub/undergrowth 0.2 0.1 0.2 5 Open/urban 0.2 0.1 0.3 In this case the agricultural land-use, coffee, is perennial and has some cavity and ground nesting resources. In a more frequently disturbed annual cropping system, nesting resources may be 0. For large monoculture cropping systems, floral resources are only available during a single crop’s blooming period, which may be as brief a period as a few weeks, and therefore not provide a very reliable resource for pollinators. It is important to consider carefully what the cropping system of interest realistically provides in the way of floral and nesting resources, because overestimating the value of cropland as a resource to pollinators will underestimate the value of natural habitat to pollinators. If different crop fields have different cropping systems and therefore different relative magnitudes of pollinator resources, it would be best to reclassify the land-use map to create a different land-use class for each cropping system. 4. Half-saturation constant (optional). The model will also prompt you to enter a half-saturation constant, which will be used when calculating the pollinator service value map. This constant converts the pollinator supply into yield and represents the abundance of pollinators required to reach 50% of pollinator-dependent yield. We suggest that the user apply the default value derived from previous work (i.e., 0.125, Lonsdorf et al 2009) unless there are data to justify changing it. The value must be greater than 0 and it is unlikely that the value would be greater than 0.2. 5. Future Scenarios (optional). To evaluate change in pollination services under a future scenario, a Future Land Cover Map needs to be provided for that future time point (along with the year depicted). The raster dataset needs to be formatted exactly like the current Land Cover Map (data input #1). This LULC map could reflect changes in land management policy, trends in land use change (e.g., agricultural expansion, urbanization, increased habitat protection). Sample data set: \InVEST\Base_Data\Terrestrial\lulc_samp_fut
19.5 Running the Model
Before running the Pollination model, make sure that the InVEST toolbox has been added to your ARCMAP docu- ment, as described in the Getting Started chapter of this guide. You will also need two additional python libraries to run the pollination model: GDAL and Numpy. The versions that you install will depend on the Python version on
19.5. Running the Model 319 InVEST User Guide, Release 2.5.6 your computer. Installation of these libraries may require you to have admin privileges on the computer. Below are the installation instructions. These instructions are for Windows XP and may differ for other versions of Windows or other operating systems: 1. Install Numpy. If you are running ArcGIS 9.3 with Python 2.5 then it is likely that Numpy is already installed. To confirm this, open Python command line from the Start menu and type ‘import numpy’ and press enter. If no error appears then Numpy is already installed. If you need to install Numpy, get the appropriate version from this location: http://sourceforge.net/projects/numpy/files/ and run the install. Ensure the version you install matches your python version. 2. Download and install GDAL from: http://download.osgeo.org/gdal/win32/1.6/gdalwin32exe160.zip. 3. Unzip the GDAL archive into a permanent location (e.g., C:\gdalwin32-1.6). 4. Add your new GDAL bin directory (C:\gdalwin32-1.6\bin, if you installed as above) to your system Path envi- ronment variable. To do this, right click on ‘My Computer,’ ‘Properties,’ Advanced > Environment Variables. Under system variables, select ‘Path’ system variable, edit, add a semicolon to separate the existing values then add your GDAL bin directory. For example if the existing Path variable was “C:\Program Files\soft,” after editing it should read “C:\Program Files\soft; C:\gdalwin32-1.6\bin” Do not delete any paths that were there before. 5. In the same Environment Variables dialog, create a new User Variable named GDAL_DATA with a value of C:\gdalwin32-1.6\data (change this to suit your GDAL install location). 6. Install the GDAL python bindings. Download the appropriate package from this location: http://pypi.python.org/pypi/GDAL/1.6.1. Browse to the bottom of that page and select a version that matches your python version. • Make sure that you have prepared the required input data files according to the specifications in Data Needs. Specifically, you will need a land cover raster file depicting the different land cover and land use types in the landscape, a Table of Land Cover Attributes, describing the suitability of the land cover types to nesting and floral resources, and a Table of Pollinator Species or Guilds, describing the nesting and seasonal behavior and crop visitation of different pollinators. • Create a workspace on your computer hard-drive if you are using your data. The pathname to the workspace should not have spaces. All your output files will be dumped here. For simplicity, you could create a folder in your workspace called “input” and place all your input files here. It is not necessary to place input files in the workspace, but this will make it easier to view the data you use to run your model. If this is your first time using InVEST and you wish to use sample data, you can use the data provided in InVEST-Setup.exe. If you unzipped the InVEST files to your C-drive (as described in the Getting Started chapter), you should see a folder called /Invest/pollination. This folder should be your workspace. The input files are in a folder called /Invest/pollination/input and in /invest/base_data. • Open an ARCMAP document to run your model. * Locate the INVEST toolbox in ARCTOOLBOX. ARC- TOOLBOX should be open in ARCMAP, but if it is not, click on the ARCTOOLBOX symbol. See the Getting Started chapter if you do not see the InVEST toolbox. • Click once on the plus sign on the left side of the InVEST toolbox to see the list of tools expand. Double-click on Pollination. • An interface will appear like the one below that indicates default file names but you can use the file buttons to browse to your data. When you place your cursor in each space, you can read a description of the data requirements in the right side of the interface. Refer to the Data Needs section for information on data formats. • Fill in data file names and values for all required prompts. Unless the space is indicated as optional, inputs are required. • After entering all required data, click OK. The script will run, and its progress will be indicated by a “Progress dialogue.” • The successful running of the model and the time it takes depends on a combination of the following factors:
19.5. Running the Model 320 InVEST User Guide, Release 2.5.6
19.5. Running the Model 321 InVEST User Guide, Release 2.5.6
• Size of landscape: If your landscape is very large (e.g., >3 million cells) then you may experience problems. Consider either entering a larger resolution than the original resolution of the image or cropping your image to a smaller extent. - Resolution: The cell size chosen for the model run determines the effective number of cells that the model has to handle. Select this carefully depending on the pollinator flight distances. - Foraging distances (Alpha): If the Alphas of the pollinators are large (>1000m) then the distance matrix becomes large, which results in a long run time or potential crashing. - Number of pollinator species: Since the model processes each pollinator in turn, the more species you have the longer it takes to complete the run. - Your computer: The memory and speed of your computer will determine the success and speed of your run. It is preferable to have at least 2GB memory and enough free disk space. - On a 3GB memory computer with a 3.5 million cells and 56m resolution, 4 pollinators with alphas between 100m and 2000m the model takes up to 3 hours to run. • Upon successful completion of the model, you will see two new folders in your workspace called “output” for final maps and “intermediate” for intermediate results. The folders should contain several raster grids, described in the next section. • Load these grids into ARCMAP using the ADD DATA button. The next section further describes what these files mean.
• To change the symbology of a layer, right-click on the layer name in the table of contents, select PROPERTIES and then SYMBOLOGY. There are many options to change the file’s appearance in the map. • To view the attribute data of output files, right click a layer and select OPEN ATTRIBUTE TABLE.
19.6 Interpreting results
19.6.1 Final results
Final results are found in the Output folder within the Workspace specified for this module. • Parameter log: Each time the model is run, a text (.txt) file will appear in the Output folder. The file will list the parameter values for that run and will be named according to the service, the date and time, and the suffix. • sup_tot_cur: This is a map of pollinator abundance index, summing over all bee species or guilds. It represents an index of the likely abundance of pollinator species nesting on each cell in the landscape, given the availability of nesting sites and of flower (food) resources nearby. • sup_tot_fut: The same as above, but for the future scenario land cover map, if provided. • frm_avg_cur: This is a map of pollinator abundance on each agricultural cell in the landscape, based on the average of all bee species or guilds. It represents the likely average abundance of pollinators visiting each farm site. • frm_avg_fut: The same as above, but for the future scenario land cover map, if provided. • sup_val_cur: This is a map of “pollinator service value”: the relative value of the pollinator “supply” in each agricultural cell to crop production in the surrounding neighborhood. It is an index derived by distributing the values in frm_val_cur (an intermediate result) back to surrounding pollinator sources, using information on flight ranges of contributing pollinators. This is a map of where pollination services are coming from, and their (relative) values. Units are not dollars per se, but the index is a relative measure of economic value. • sup_val_fut: The same as above, but for future scenario land cover map, if provided.
Intermediate results
You may also want to examine the intermediate results. These files can help determine the reasons for the patterns in the final results. They are found in the Intermediate folder within the Workspace specified for this module.
19.6. Interpreting results 322 InVEST User Guide, Release 2.5.6
• hn_
19.7 Appendix: Data sources
List of globally important crops and their dependence on animal pollinators: (Klein et al. 2007).
19.8 References
Allen-Wardell, G., P. Bernhardt, R. Bitner, A. Burquez, S. Buchmann, J. Cane, PA Cox, V. Dalton, P. Feinsinger, M. Ingram, D. Inouye, CE Jones, K. Kennedy, P. Kevan, and H. Koopowitz. 1998. The potential consequences of pollinator declines on the conservation of biodiversity and stability of food crop yields. Conservation Biology 12: 8-17. Cane, JH. 1997. Lifetime monetary value of individual pollinators: the bee habropoda laboriosa at rabbiteye blueberry (vaccinium ashei reade). Acta Horticulturae 446: 67-70. Costanza, R., R. d’Arge, R. de Groot, S. Farber, M. Grasso, B. Hannon, K. Limburg, S. Naeem, RV O’Neill, J. Paruelo, RG Raskin, P. Sutton, and M. van den Belt. 1997. The value of the world’s ecosystem services and natural capital. Nature 387: 253-260. Free, JB. 1993. Insect pollination of crops. Academic Press, London.
19.7. Appendix: Data sources 323 InVEST User Guide, Release 2.5.6
Greenleaf, SS, NM Williams, R. Winfree, and C. Kremen. 2007. Bee foraging ranges and their relationship to body size. Oecologia 153: 589-596. Greenleaf, SS, and C. Kremen. 2006. Wild bee species increase tomato production and respond differently to sur- rounding land use in Northern California. Biological Conservation 133:81-87. Klein, AM, BE Vaissiere, JH Cane, I. Steffan-Dewenter, SA Cunningham, C. Kremen, and T. Tscharntke. 2007. Importance of pollinators in changing landscapes for world crops. Proceedings of the Royal Society B-Biological Sciences 274: 303-313. Kremen, C., NM Williams, RL Bugg, JP Fay, and RW Thorp. 2004. The area requirements of an ecosystem service: crop pollination by native bee communities in California. Ecology Letters 7: 1109-1119. Lonsdorf, E., C. Kremen, T. Ricketts, R. Winfree, N. Williams, and SS Greenleaf. 2009. Modelling pollination services across agricultural landscapes. Annals of Botany 1: 12 online [http://aob.oxfordjournals.org/cgi/content/abstract/103/9/1589]. Lonsdorf, E., TH Ricketts, CM Kremen, NM Williams, and S. Greenleaf. in press. Pollination services in P. Kareiva, TH Ricketts, GC Daily, H. Tallis, and S. Polasky, eds. The theory and practice of ecosystem service valuation. Losey, JE, and M. Vaughan. 2006. The economic value of ecological services provided by insects. Bioscience 56: 311-323. Nabhan, GP, and SL Buchmann. 1997. Services provided by pollinators. Pages 133-150 in GC Daily, ed. Nature’s services. Island Press, Washington, D.C. Priess, JA, M. Mimler, AM Klein, S. Schwarze, T. Tscharntke, and I. Steffan-Dewenter. 2007. Linking deforestation scenarios to pollination services and economic returns in coffee agroforestry systems. Ecological Applications 17: 407-417. Ricketts, TH. 2004. Tropical forest fragments enhance pollinator activity in nearby coffee crops. Conservation Biology 18: 1262-1271. Ricketts, TH, NM Williams, and MM Mayfield. 2006. Connectivity and ecosystem services: crop pollination in agricultural landscapes. Pages 255-289 in M. Sanjayan and K. Crooks, eds. Connectivity for Conservation. Cambridge University Press, Cambridge, UK. Roubik, DW, and M. Aluja. 1983. Flight ranges of Melipona and Trigona in tropical forest. Journal of the Kansas Entomological Society 56: 217-222. Southwick, EE, and L. Southwick. 1992. Estimating the economic value of honey-bees (Hymenoptera; Apidae) as agricultural pollinators in the United States. Journal of Economic Entomology 85: 621-633. Winfree, R., J. Dushoff, EE Crone, CB Schultz, RV Budny, NM Williams, and C. Kremen. 2005. Testing simple indices of habitat proximity. American Naturalist 165(6): 707-717.
19.8. References 324 Part IV
Experimental Models
325 CHAPTER TWENTY
MALARIA MOSQUITO HABITAT
20.1 Summary
Model the suitability of mosquito breeding areas, the impact on a human population, and the relative impacts of the landscape.
20.2 Introduction
20.3 The model
20.3.1 How it works
20.3.2 Mosquito breeding site suitability
The suitability index of a mosquito breeding ground in a given pixel is defined as
−si qi = bie log fi where
• qi is the suitability rating of cell i
• bi is the land type breeding suitability rating of the land cover in cell i, this is a function of the land cover type in cell i
• si the slope of cell i
• fi the flow accumulation at cell i.
20.3.3 Breeding Site Influence
Once the suitability of a mosquito breeding grounds is calculated for a landscape, the influence of that ground over a population center in cell i given as
X −dij α ci = pi qje j∈landscape where
326 InVEST User Guide, Release 2.5.6
• dij is the distance from the center of cell i to cell j
• pj is the human population density of cell j • α is the mean flight distance of the mosquito species
20.3.4 Influential Breeding Sites
Once the influence of the breeding grounds are projected on a landscape, the most influential breeding sites can be calculated as
X −dij fi = qi cje j∈landscape
20.4 Limitations and simplifications hi
20.5 Data Needs
• Workspace: The directory to hold output and intermediate results of the particular model run. After the model run is completed the output will be located in this directory. To run multiple scenarios, create a new workspace for each scenario. • DEM: A height map raster of the area used to calculate slope and flow accumulation. • Land cover map: A raster indicating land types which is used to determine base breeding ground suitability. • Land cover type suitability table: a csv file with the fields lucode corresponding to the land use codes in the land cover map and breeding_suitability_index which is a number between 0 (not suitable) and 1 (most suitable) indicating the suitability of that land cover type for mosquito breeding grounds. • Population Density map: A raster indicating human population density. • Mean mosquito flight distance: The mean distance in meters that a mosquoto will travel to feed from its breeding site. • Threshold Flow Accumulation: The number of upstream pixels before a cell is classified as a running stream (not standing water). • Amount of area available for conversion: A value in hectares of how much land is under consideration for conversion. The model will indicate the area on the land cover map most suitable for conversion to mitigate influence from mosquito breeding grounds.
20.6 Running the Model
If you encounter any errors please email the output log to [email protected].
20.4. Limitations and simplifications 327 InVEST User Guide, Release 2.5.6
20.7 Interpreting Results
20.7.1 Model Outputs
Each of model output files is saved in the Output and Intermediate folders that are saved within the user- specified workspace. Output folder • Output\\breeding_suitability.tif: The breeding suitability index, a function of land cover, slope, and flow accumulation. • Output\\breeding_site_influence.tif: These are the areas that humans live that are most influ- enced by mosquito breeding grounds. • Output\\influential_breeding_site.tif: The values in this raster indicate the relative intensity of the influence of a breeding ground on the human population. • Output\\optimal_conversion.tif: This output exists of land use optimization was selected. It indi- cates the areas with the most marginal gain in malaria avoidance if converted to a land cover type not suitable for mosquito breeding.
20.8 References
Packard, R. M. 2007. The Making of a Tropical Disease.
20.7. Interpreting Results 328 Part V
Acknowledgements
329 CHAPTER TWENTYONE
MARINE MODELS
21.1 Acknowledgements
21.1.1 Data sources
The Marine InVEST development team would like to acknowledge the following sources for data that are provided with the models: WAVEWATCH III model hindcast reanalysis results are from NOAA’s National Weather Service ETOPO1 was developed by and is available from NOAA’s National Geophysical Data Center (NGDC). (Amante, C. and B. W. Eakins, ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Technical Memorandum NESDIS NGDC-24, 19 pp, March 2009). The Global Self-consistent, Hierarchical, High-resolution Shoreline Database (GSHHS) is developed and main- tained by Paul Wessel, SOEST, University of Hawai’i, Honolulu, and Walter H.F. Smith, NOAA Geosciences Lab, Na- tional Ocean Service, Silver Spring, MD. It can be accessed via NOAA’s National Geophysical Data Center (NGDC). (Wessel, P., and W. H. F. Smith, A Global Self-consistent, Hierarchical, High-resolution Shoreline Database, J. Geo- phys. Res., 101) British Columbia Shorezone Data are provided courtesy of the Province of British Columbia, Ministry of Natural Resource Operations, GeoBC Division. The data used for this model is a snapshot in time. For the most current coastal resource and shorezone data please visit http://www.geobc.gov.bc.ca. Gridded Population of the World Version 3 (GPWv3) data are provided courtesy of the Center for Interna- tional Earth Science Information Network (CIESIN), Columbia University; and Centro Internacional de Agricul- tura Tropical (CIAT). 2005. Gridded Population of the World Version 3 (GPWv3): Population Grids. Pal- isades, NY: Socioeconomic Data and Applications Center (SEDAC), CIESIN, Columbia University. Available at http://sedac.ciesin.columbia.edu/gpw. Department of Fisheries and Oceans Canada provides commercial fishery data layers (groundfish trawl and long- line, salmon troll, and shrimp trawl) collected from 1993-95 through interviews with fisheries officers for the West Coast of Vancouver Island, British Columbia. Data are publicly available through GeoBC, the geographic information clearinghouse for British Columbia Canada. Habitat layers are available through British Columbia Marine Conservation Analysis (BCMCA) and GeoBC.
21.1.2 Individuals and organizations
We would also like to thank the following individuals and organizations for support, guidance, great collaborations, and fun! This is not an exhaustive list. The Gordon and Betty Moore Foundation
330 InVEST User Guide, Release 2.5.6
The West Coast Aquatic Management Board NOAA Fisheries, Northwest Fisheries Science Center Gretchen Daily, Peter Kareiva, Taylor Ricketts, Steve Polasky, and Jon Foley Our Technical Working Group: Mary Ruckelshaus, Anne Guerry, Katie Arkema, Greg Guannel, CK Kim, Mike Papenfus, Jodie Toft, Gregg Verutes, Joey Bernhardt, Apollo Qi, Jeremy Davies, Heather Tallis, Steve Polasky, Mark Plummer, Phil Levin, Bill Labiosa, Francis Chan, Guy Gelfenbaum, Peter Ruggiero, Andre Punt, Melanie McField, Ben Halpern, Sarah Lester, Malin Pinsky, Mike Beck, Barry Gold, Kai Chan Others: Matt Marsik, Spencer Wood, Dave Sutherland, Andrew Day, Laura Loucks, Trudy Warner, Kevin Head, Roger Bedard, Jim Regetz, Dan Holland, Jameal Samhouri
21.1. Acknowledgements 331 Part VI
Roadmap
332 CHAPTER TWENTYTWO
MARINE MODELS
22.1 What’s coming up for Marine InVEST models?
22.1.1 Planned modifications to InVEST models
Wave energy
(Tier 1) The goal of the wave energy model is to provide planners with wave energy siting information for growing energy demand. The wave energy model calculates potential wave power and wave energy that can be captured by currently available technology. The model evaluates the economic feasibility of potential wave energy conversion facilities using a cost-benefit analysis. The spatially explicit model outputs allow users to examine potential conflicts with other uses of the marine environment (e.g. fishing, shipping, oil rigs, etc) that co-occur with regions suitable for wave power plants. Thus, the model helps decision-makers understand where best to install a facility while balancing desires for the greatest captured wave energy and the least effect on other human uses in coastal and ocean ecosystems. What’s coming up in future releases? • Incorporating local wave input data: The current version includes global and regional wave information as base model input, which allow first approximation of wave energy anywhere in the world ocean. In future releases, the model will allow users to provide their own wave input data if desired. • Valuing economic feasibility at a global and local scale: The current version allows economic valuation of a wave energy conversion facility at a local scale. In future releases, the model will allow users to evaluate wave energy projects at global and regional scales using global ports data.
Coastal Vulnerability
(Tier 0) The Marine InVEST Coastal Vulnerability model maps the location of people living in coastal areas and the relative exposure of coastal communities and environments to erosion and flooding during large storms. In particular, it highlights the shoreline protection services provided by natural habitats. Outputs from the model can be used in a number of ways including: helping to understand what factors (natural and social) make a stretch of coastline more or less vulnerable to erosion and flooding, estimating the change in exposure that might result from a planned management action, and informing where and how coastal development might occur. The Coastal Vulnerability Model can also be run to help identify regions where the more quantitative (Tier 1) Marine InVEST Coastal Protection model might best be applied.
333 InVEST User Guide, Release 2.5.6
Limitations of current model
• Wind fetch distances are only computed for 16 directions, separated by 22.5 degrees • Wind speed values used to estimate wind exposure and characteristics of wind-generated waves are statistics of highest wind speeds, not raw wind speed values • Maximum fetch distance computed by the model is 50km. This is also the threshold to determine whether a shoreline segment is exposed to the open ocean. • We only use one value of water depth to compute wind-generated waves for the whole AOI. • Surge potential of a sheltered shoreline segment is the same as the surge potential of the closest exposed segment. • Effect of structures is modeled by decreasing the ranking of a cell adjacent to a structure by 1 point. We do not estimate the length of influence of the structure. We do not model its impacts on subtidal environment. • Map of tidal range is not included
What’s coming up in future releases?
• Inclusion of more social indicators to link vulnerability to populations • Simple valuation methods • Allow users to define regions that are shallow in the area of interest. • Inclusion of tidal range maps
Coastal Protection
(Tier 1) The InVEST Coastal Protection model quantifies the protective benefits that natural habitats provide against erosion and inundation in nearshore environments. It is composed of two sub-models: a Profile Generator and a Nearshore Wave and Erosion model. In the absence of local data detailing the profile of the shoreline, the Profile Generator model helps users combine information about the local bathymetry and backshore to generate a 1-Dimensional (1D) cross-shore (perpendicular to the shoreline) beach profile. The Nearshore Waves and Erosion model uses the shoreline profile to compute summaries of nearshore wave information and outputs the total water level and the amount of shoreline erosion in the presence and absence of nearshore marine habitats (e.g., coral or oyster reefs, vegetation, sand dunes).
Limitations of current model
• It is a 1D model based on linear wave theory. We ignore any wave processes that occur over a complex bathymetry. We also ignore any non-linear wave processes, especially when they encounter natural habitats. • We model wave attenuation over coral and oyster reef in a simple way, using empirical methods. • We assume that habitats are not affected (e.g., uprooted or broken) by a storm. Also, we assume that they have a constant friction and drag coefficient that is independent of the level of turbulence in the water column. • We use simple beach and consolidated sediment erosion models that ignore any dynamic feedback between waves and the seabed.
22.1. What’s coming up for Marine InVEST models? 334 InVEST User Guide, Release 2.5.6
What’s coming up in future releases?
• Valuation (in social and economic currencies) of the role of natural habitats. • Guidance on the modeling of marsh profiles. • Estimation of marsh edge erosion. • Better estimation of erosion of muddy substrates. • Inclusion of diffration effects in estimation of wave heights leeward of oyster reefs. • Inclusion of structures in the model and indication of their impacts on subtidal habitats and adjacent properties, as well as quantification of the amount overtopping that occurs during storms. • Inclusion of more friction and drag coefficient that are a function of turbulence level in the water column. • Option to batch process Nearshore Wave and Erosion model to run it for multiple locations at once. • Visual representation of outputs of the Nearshore Wave and Erosion model in map form. • Inclusion of stand-alone tools to model wave evolution in the presence or absence of specific natural habitats. • Inclusion of wind-wave generation tool to estimate wave height and wave period based on fetch distance and wind speed.
Finfish Aquaculture
(Tier 1) The current version of the InVEST aquaculture model analyzes the production of farmed Atlantic salmon (Salmo salar) based on farming practices, water temperature and economic factors. While the current model quantifies the economic benefits of salmon aquaculture, it does not quantify the associated environmental costs. Future versions of the aquaculture model will allow users to quantify the environmental impacts of salmon aquaculture, including the release of farm wastes and the transfer of disease vectors from farmed to wild populations. These impacts may have effects on other environmental services, such as water quality and wild salmon fisheries. When combined with the full suite of InVEST model outputs, this added functionality will allow InVEST users to explore the environmental costs and benefits of Atlantic salmon production, as well as subsequent effects on other environmental services. In addition, subsequent releases will allow users to model production of other finfish species and shellfish (see “Shellfish Production” in the “Additional Models Under Development” section).
What’s coming up in future releases?
• Quantification of farm waste production. Outputs will include dissolved and particulate wastes generated as a byproduct of Atlantic salmon production. These outputs will be available for use in other InVEST models (e.g., water quality, habitat quality, fisheries) to assess impacts of Atlantic salmon aquaculture on other environmental services. • Quantification of impacts of parasitic sea-lice. Disease outbreaks and parasitism cause substantial financial burdens to aquaculturists (e.g., due to costs of treatment, foregone revenue due to culling of infected fish). In addition, high stocking densities of farmed fish increase the risk of transmission of sea-lice from farmed to wild salmon. This sub-module will allow users to quantify the impact of parasitic sea-lice on farmed salmon production and the risk of disease transmission to wild salmon populations. • Guidance on modifying the Atlantic salmon model for other cultured marine finfish. • A shellfish aquaculture model (see “Shellfish Production” in the “Additional Models Under Development” sec- tion). This model will quantify volume, economic value, filtration and production of wastes of cultured shellfish (e.g., oyster, shrimp).
22.1. What’s coming up for Marine InVEST models? 335 InVEST User Guide, Release 2.5.6
• Changes in time steps. The current model operates at a daily time step (requiring daily temperature data), but future iterations will allow for monthly or yearly temperature inputs.
Aesthetic Quality
(Tier 0) This model examines the production of aesthetic views by the coastal and marine environment. It can be used to examine how views are likely to be impacted by various marine and coastal activities (e.g. wave energy facilities, aquaculture facilities, coastal clear cuts, coastal development, mining, etc).
What’s coming up in future releases?
• Additional functionality to analyze aesthetics from a viewer’s perspective. The tool will offer viewshed output maps for up to 20 different locations. • Incorporation of visual quality ratings for various land use/land cover (LULC) types • Enabling users to map relative changes in aesthetic quality across multiple scenarios • Incorporation of global forest cover dataset to control for areas that are not visible because forest cover • Provision of higher resolution topographic base data (SRTM 3-90m resolution)
Marine Habitat Risk Assessment
(Tier 0) The InVEST Habitat Risk Assessment model allows users to assess the risk posed to coastal and marine habitats by human activities and the potential consequences of exposure to that risk for the delivery of environmental services and biodiversity. Outputs from the model are useful for understanding the relative risk of human activities and climate change on habitats within a study region under alternative future scenarios and for identifying which habitats are of high enough quality to provide the services people care about.
What’s coming up in future releases?
• Connectivity scoring: The current model allows users to score connectivity of habitat patches based on known dispersal distances. In future releases, the model will calculate the distance to the nearest neighboring habitat patch to improve the connectivity scoring. • Decay exposure: The current model assumes that exposure to stressors is uniform in space within a stressor’s user-defined zone of influence. In future releases, the model will allow for the decay of the degree of exposure from the center of a stressor’s location to the outer edge of its zone of influence. • Spatial variation in intensity scores: In the current version of the model the intensity of each stressor is scored for the whole study region. In future versions of the model, the user will be able to input a stressor intensity layer so that stressor intensity can vary within the study region.
Overlap Analysis: Fisheries (Tier O) and Recreation (Tier O)
(Tier 0) The InVEST Overlap Analysis Model was designed to produce maps that can be used to identify marine and coastal areas that are most important for human use. The model produces a map of hotspots for human activities (e.g., fishing activity/fishing grounds, various recreational activities) across as many human uses as the users chooses to
22.1. What’s coming up for Marine InVEST models? 336 InVEST User Guide, Release 2.5.6 include. Outputs can be used to help decision-makers weigh potential conflicts between sectors of spatially-explicit management options that may involve new activities or infrastructure.
What’s coming up in future releases?
• Functionality for raster inputs: The current model accepts point and polygon data but not raster inputs. We anticipate that users working with fisheries data, in particular, may have raster data that they would like to use as model inputs. Future releases will accommodate the use of raster data. • Output options - absolute values: The current model calculates an Importance Score based on inputs about where human uses occur and, optionally, their relative weighting. Users can base these weights on a variety of metrics, including catch and revenue, but the outputs are still scaled to a score, not an absolute value. In future releases, users will have the option to output Importance Scores and absolute values.
22.1.2 Additional models under development
Fisheries production
Simple population models
(Tier 1) A generic and flexible model will be included in InVEST for estimating the quantity and value of fish harvested by commercial fisheries. It will be appropriate for use with single species or groups of species. For example, one could choose to parameterize it for each of the top 3 commercially important species in a region. In its most general form, the model estimates the annual production of fish, which is the biomass in the previous year multiplied by a function that captures changes to habitat, fishing or from climate change. The model incorporates the impacts of biogenic habitat on the survival and fecundity of different life- stages of target species. We use a matrix structure to transition fish from one year to the next. The matrix is parameterized using information on stage (or age) specific survival, fecundity and harvest. Scenarios are represented as changes to fecundity, survival and harvest rate. It is possible to adjust the temporal duration of the model runs according to the life-history characteristics of the species being modeled (e.g., fast or slow turnover species). Individual models are in development for spiny lobster in Belize, Pacific salmon in British Columbia and Dungeness crab in Washington State, and are being tested for those species in other geographies and generalized for species with similar life histories.
Inputs • Spatial delineation of areas to include in the model • Age- and area-specific estimates of number of adults • Sector-specific catches or catch-per-unit-effort, by area if available • Harvest rates (fixed or variable) • Proportion of age-specific returns • Productivity (e.g. pre-smolt/spawner; recruit/spawner) • Survival estimates (age- or stage-specific and/or area-specific) • Table describing influence of human activity or climate change on survival and productivity. • Cost and price information for the commercial sector
22.1. What’s coming up for Marine InVEST models? 337 InVEST User Guide, Release 2.5.6
Outputs • Catch (area-specific if data exist) partitioned by sector (e.g. commercial and recreational). • Value of commercial catch
Future features
Future versions of InVEST will facilitate the use of more complex food-web models (e.g., Ecopath with Ecosim and Atlantis). The InVEST interface will communicate these more complex models outside of InVEST so that parameters can be compared to outputs from other InVEST models such as aquaculture, wave energy, and coastal protection.
Recreational fishing sub-model
(Tier 1) The Marine InVEST recreational fishing submodel allows users to evaluate the recreational benefits of improving management of marine and coastal fisheries. This model is directly linked to the Marine InVEST fisheries production model and is designed to reflect changes in the output of the fisheries production model as well as simple changes in the management of the recreational sector (e.g. allocation of more fish to the recreational sector). In the event that a scenario indicates increases in fish abundance available for recreational harvest, this model predicts the amount of recreational fishing effort required to catch those additional fish. This can be thought of as a proxy measure for the potential supply of recreational opportunity in the fisheries sector. With additional local data, this potential supply of fishing opportunity can be translated into potential benefits that could be realized by recreational anglers, recreational fishing operators, and a proxy measure of expenditures in the local economy.
Inputs
• Change in fish abundance (predicted by InVEST fisheries model) • Estimate of catch per unit effort for each target species in the recreational sector • Coefficient that determines the number of fish that will be available for catch by recreational anglers. This parameter can be changed to reflect alternative management scenarios or it can be based on current management policy. • Site-appropriate estimate of consumer surplus per day of fishing • Site-appropriate measure of angler expenditures per trip. • Site-appropriate measures of operating costs (fuel costs, labor costs, docking, etc.).
Outputs
The primary output from the recreational fishing sub-model is an estimate of the amount of effort (number of trips) that would be required to catch the number of fish allocated to the recreational sector as predicted by the InVEST fisheries production model. The estimated effort required to catch these fish represents an upper bound on effort and could be much lower given constraints on fleet capacity and the number of anglers visiting the area to participate in recreational fishing. In additional to estimating an upper bound on fishing effort, the model also generates estimates of the economic benefits accruing to: • Recreational anglers (Consumer surplus per trip scaled by the number of trips). • Recreational fishing operators (net revenue per trip) • Expenditures introduced to the local economy
22.1. What’s coming up for Marine InVEST models? 338 InVEST User Guide, Release 2.5.6
Water quality
(Tier 1) Although water quality is not an environmental service per se, it is an important intermediate output that can connect other InVEST models. The water quality (WQ) model simulates the movement and fate of water quality variables (state-variables) in response to changes in ecosystem structure driven by various management decisions and human activities. Hence, this model assesses how management and human activities influence the water quality in coastal and estuarine ecosystems. The model can be used for diagnosing the type of WQ problems (e.g., hypoxia, eutrophication, high concentrations of bacteria and toxic chemicals) expected, identifying environmental control aspects for water quality, and setting water quality standards. The WQ model can be linked with other Marine InVEST models to evaluate environmental services relating to fisheries, aquaculture, habitat quality, and recreation. Consequently, the WQ model can help decision-makers establish management strategies for the desirable use of a water body.
WQ Model Tier 1a
Initial development is underway of a simplified physical transport model that will give decision-makers a qualitative assessment of where water quality issues may arise in an estuarine system. The model will output residence time, which when coupled with river and nutrient inputs, will allow a general look at where water quality issues such as hypoxia or eutrophication may occur. The model will be based on a one- or two- dimensional finite segment configuration (the choice is set by the characteristics of the estuarine system) that incorporates physical transport processes driven by river discharge and tidal dispersion. The model will simulate mass transport along the main channel of a system.
WQ Model Tier 1b
The second development underway will tie the water quality variables (e.g., nutrients or dissolved oxygen) to the physical transport model. Although the targeted time scale is monthly to seasonal, we will first produce annual-average distributions of water quality state variables. Box modeling approaches are also being considered to accommodate more flexible applications across multiple scales in coastal and estuarine systems, which may be appropriate for data rich areas.
Inputs The WQ model requires: • Estuarine coefficient tables – Geomorphology (e.g., depth, width, and length of an estuarine system) – River discharge input at the upstream boundary – Tidal dispersion coefficient, which can be estimated using salinity distribution. We will also provide a lookup table or an empirical equation using tidal strength to estimate tidal dispersion coefficient in places with limited data. • WQ state variables (e.g., nutrients, metals, viruses, toxic chemicals, dissolved oxygen, etc.). Nitrogen and Phosphorus would be the first target variables. • Loading – Point sources, loading from discharge pipes, sewage treatment outfall, aquaculture farms, etc. – Non-point sources, loading from agricultural, urban and suburban runoff, groundwater, etc. – Watershed models can be used to estimate both point and non-point source loading from land. • Kinetic coefficients
22.1. What’s coming up for Marine InVEST models? 339 InVEST User Guide, Release 2.5.6
Outputs The WQ Model Tier 1a: • Assesses areas in an estuary that are at-risk to water quality issues The WQ Model Tier 1b: • Produces spatially explicit concentration maps of water quality state variables • Evaluates watershed/coastal management strategies to maintain desirable water quality standards
Carbon storage and sequestration
(Tier 1) Marine and terrestrial ecosystems help regulate Earth’s climate by adding and removing greenhouse gases (GHGs) such as carbon dioxide (CO2) from the atmosphere. Coastal marine plants such as mangroves and seagrasses store large amounts of carbon in their sediments, leaves and other biomass. By storing carbon in their standing stocks, marine ecosystems keep CO2 out of the atmosphere, where it would otherwise contribute to climate change. In addition to storing carbon, marine ecosystems accumulate carbon in their sediments continually, creating large reser- voirs of long-term carbon sequestration. Management strategies that change the cover of marine vegetation, such as seagrass restoration or mangrove clearing, can change carbon storage and the potential for carbon sequestration on seascape. The InVEST Carbon Model estimates how much carbon is stored in coastal vegetation, how much carbon is sequestered in the sediments, and the economic value of storage and sequestration. The approach is very similar to that of the terrestrial carbon model.
Inputs
• Maps of the distribution of nearshore marine vegetation (i.e. mangroves, salt marshes) • Data on the amount of carbon stored in four carbon ‘pools’: aboveground biomass, belowground biomass, sediments, and dead organic matter. • Data on the rate of carbon accumulation in the sediments for each type of marine vegetation. • Additional data on the market or social value of sequestered carbon and its annual rate of change, and a discount rate can be used in an optional model that estimates the value of this environmental service to society.
Outputs
• Carbon storage (Mg C/ha). • Carbon sequestration (Mg C/ha/yr). • Economic value of carbon storage and sequestration.
Shellfish Production (wild and aquacultured)
(Tier 1) In this model, we map how incremental changes in ecosystem structure (e.g., water quality attributes including temper- ature, salinity, nutrient availability), changes to wild harvest or changes to operations at specific aquaculture facilities affect shellfish production and commercial value and nutrient filtration. We use a framework similar to the Farm Aqua- culture Resource Management model 1, which has been developed for assessment of individual coastal and offshore
1 10. (a) Ferreira, A.J.S. Hawkins, S.B. Bricker, 2007. Management of productivity, environmental effects and profitability of shellfish aqua- culture – the Farm Aquaculture Resource Management (FARM) model. Aquaculture, 264, 160-174.
22.1. What’s coming up for Marine InVEST models? 340 InVEST User Guide, Release 2.5.6 shellfish aquaculture farms. The model contains two linked sub-models that represent 1) shellfish individual growth and 2) shellfish population dynamics.
Inputs
• shellfish growing area dimension (e.g., width, length, depth – if suspended) • shellfish cultivation (e.g., species, size at outplanting for aquaculture or settlement for wild stock, target harvest size or weight, density of individuals) • (optional for commercial valuation) product (e.g., half shell or shucked oyster), market price of product, har- vesting and processing costs, facility operation costs (aquaculture facilities) • environmental variables (e.g., water temperature and current speed; available food as concentration of Chloro- phyll a, dissolved inorganic nitrogen - DIN, particulate organic matter - POM); these can be from local data or, when available, from the InVEST Water Quality models.
Outputs
• harvestable biomass • harvestable number of animals • filtration of Chlorophyll a, DIN and POM • (optional for commercial valuation) net revenue of product
Recreation
(Tier 1) The forthcoming InVEST recreation model predicts where people go to recreate (or, more specifically, the spread of person-days of recreation in space). It does this using attributes of places, such as natural features (e.g., parks), built features (e.g., roads) and human uses (e.g., industrial activities) among others. Because these attributes are often good predictors of visitation rates, the recreation model will come pre-loaded with data about these and other attributes that are linked to attractiveness. We will also allow users to upload their own spatial data reflecting additional attributes that might be correlated with people’s decisions about where to recreate. Then, armed with these estimates, users will be able to use the model to predict how future changes to the landscape will alter visitation rates. The tool will output maps showing current patterns of recreational use and, optionally, maps of future use under different scenarios.
Inputs
• area of interest
Outputs
• person-days of recreation in grid cells within the area of interest
22.1. What’s coming up for Marine InVEST models? 341